Nineteenth Anniversary Issue RCA Engineer Staff W.O

Home , RCA Records

www.americanradiohistory.com Nineteenth anniversary issue RCA Engineer Staff W.O. Hadlock Editor J.C. Phillips Associate Editor With the exception of its annual anniversary issue, the RCA Engineer develops J.P. Dunn Art Editor each issue around a central technological theme, with a number of articles Diane Ahearn Editorial Secretary coming from the RCA organizations primarily concerned with that specific Juli Clifton Subscriptions technology. P.A. Gibson Composition

Since the anniversary issue includes contributions from most divisions and on a Consulting Editors broad range of subjects, it seems fitting that RCA Laboratories has been asked the RCA to provide its foreword. As the central research organization for C.A. Meyer Technical Publications Adm., Corporation, we at RCA Laboratories feel that we must cover the entire Electronic Components electronics spectrum, rather than being limited to a few specific themes, to C.W. Sall Technical Publications Adm., accomplish our objective. Laboratories F.J. Strobl Technical Publications Adm., Stated simply, that objective is the development of new RCA products and Technical Communications services plus assisting the RCA product divisions in improving existing products and services. Editorial Advisory Board Obviously, we can achieve our objective only when we have good two -way communications with the product divisions. And the RCA Engineer is one of the R.M. Cohen Dir., Quality and Reliability Laboratories' and the Corporation's communications tools. Assurance, Solid State Div. F.L. Flemming VP, Engineering, NBC As the RCA Engineer starts its 20th year with this issue, Bill Hadlock, the Editor, Television Network Div. and his colleagues are to be congratulated for the fine work they have done in C.C. Foster Mgr., Scientific Publications RCA Laboratories providing an outstanding channel for RCA's scientists and engineers to Products exchange ideas. Although it probably has not eliminated the process, the M.G. Gander Manager, Consumer Adm., RCA Service Co. Engineer certainly has reduced the number of times RCA engineers have "re- invented the wheel "! W.R. Isom Chief Engineer. RCA Records In recent years, we feel that two -way communications between the Laboratories L.R. Kirkwood Dir., Color TV Engineering and Strategic Planning, and the RCA product divisions have been improved. Yet, I cannot let this Consumer Electronics opportunity pass without acknowledging that they could be better. Therefore, I'm inviting readers of the RCA Engineer to send me their suggestions on how C.H. Lane Div. VP, Technical Planning Electronic Components the Laboratories can better serve the product divisions. Reasoned letters on this subject will be beneficial to the Laboratories, the RCA product divisions, and the H. Rosenthal Staff VP, Engineering Corporation. Research and Engineering P.Schneider Exec. VP, Leased Facilities and Engineering RCA Global Communications, Inc. /11-m Dr. W.J. Underwood Manager, Engineering Professional Programs William M. Webster Dr. H.J. Woll Division VP, Vice President Government Engineering RCA Laboratories Princeton, N.J.

Our cover

.symbolizes the improved performance delivered by transistorized ignition systems developed by RCA. The power transistor in the foreground is used on most 1974 Chrysler Corporation cars. The engineers in the photo - Wil Bennett (right) and Gil Lang - are two members of the team that won a 1974 David Sarnoff Award for Outstanding Technical Achievement for "development of high voltage power transistors for automotive ignition and other applications."

Cover concept: Joan Dunn. Photography: John Semonish, Electronic Components, Clark. NJ. Automobile courtesy of Liccardi Motors, Inc.; Route 22, Greenbrook, NJ.

www.americanradiohistory.com Vol. 201 No. 1 Junl Jul 1974 alCBM Engineer A technical journal published by To disseminate to RCA engineers technical informa- help publicize engineering achievements in a manner RCA Research and Engineering tion of professional value To publish in an appropri- that will promote the interests and reputation of RCA in Cherry Hill, N.J. ate manner important technical developments at RCA. the engineering field To provide a convenient means Bldg. 204 -2 (PY -4254) and the role of the engineer To serve as a medium of by which the RCA engineer may review his professional interchange of technical information between various work before associates and engineering RCA Engineer articles are indexed groups at RCA To create a community of engineer- management To announce outstanding and unusual annually in the April -May issue and in ing interest within the company by stressing the in- achievements to RCA engineers in a manner most likely the Index to RCA Technical Papers. terrelated nature of all technical contributions To to enhance their prestige and professional status.

Contents Emphasis- Nineteenth anniversary Editorial input ex- RCA'ers - an update F.J. Strobl 2

Engineer and the Corporation History and development of the color picture tube E.W. Herold 4

Cover feature High -voltage silicon devices open doors to electronic ignition J.S. Varal M.S. Reutter 10 General interest Surface acoustic -wave filter for television intermediate frequencies Dr. J. A. van Raalte 15

Very high resolution radiometer A.I. Aronson 20

High- capacity high- data -rate instrumentation tape recorder system O.E. Bessette 26

Linear transistor power amplifiers Dr. E.F. Belohoubekl D.S. Jacobson 29

Digital image transform encoding W. B. Schaming 32

Numerical control applications for spacecraft components D. Charrierl M. Luft 38

Data acquisition, control, and display for Navy ORTS T.Taylorl M. LeVarn 41

Low -cost distributed -port reflex loudspeaker enclosure C.C. Rearick 48

900 -MHz and 450 -MHz mobile radio performance in urban hilly terrain F.A. Barton) G.A. Wagner 52

Computer program architectural design for weapon system radar control T.H. Mehling 58

Toll switching plan for Alaska V. Batral E.F. McGill 65

TTUE -4A uhf television exciter J.B. Bullock 68

Laser scanning as a tool for testing integrated circuits Dr. R. Williams) P.V. Goedertierl Dr. J.D. Knox 76 Charge -coupled imager for 525 -line television R.L. Rodgers 79

Power switching using solid -state relays T.C. McNulty 83

Engineering and Adhesive bonding of semiconductors to substrates R.D. Larrabee 87 Research Notes Millivolt source for temperature programming of laboratory furnaces R. Fehlmanns A. E. Widmer 87

Department Pen and Podium 89

Patents Granted 91

News and Highlights 92

www.americanradiohistory.com ex-RCA'ers editorial input an update

F. J. Strobl*

Far !el t: Ted Smith, unofficial president of the ex- RCA'ers welcomes the members to AED and expresses ap- preciation to his hosts at AED for the a fine reception extended to the group. Over year ago, we reported on "the Left: C. S. Constantino, Division Vice ex- RCA'ers " -an organization of President and General Manager, of AED welcomed the ex- RCA'ers, and RCA retirees who gather monthly to many former associates among them, chat and reminisce. And we noted to the Space Center and gave a slide presentation of the facility's history. that this Delaware Valley group is Below left: After lunch in the AED unique in that it has no official name, cafeteria, the ex- RCA'ers listen to a briefing of the projects and activities at by -laws, charter, elected officers, the RCA Space Center. Below: Barton Kreuzer, former Division dues, or formal agenda. Vice President and General Manager of the Astro- Electronics Division (1960- 1967: and recently retired as Executive Today, these "non- organizational at- Vice President, is introduced as one of tributes" remain unchanged. But the the new, distinguished ex- RCA'ers. size of the membership roster has jumped from 150 (at our last report) to more than 300 members, collectively representing past ties to almost every RCA activity. Through the various stages of their growth, the ex- RCA'ers were forced to find larger and larger facilities for their monthly meetings. A banquet room at the Mallard Inn in Mount Laurel, N.J., presently serves as their rendezvous point.

Meetings are held the second Mon- day of each month (except July and August). When meeting at the Mallard Inn, the group assembles at 11:30 a.m. for some preliminary socializing before lunch. Luncheons are paid for in the club's traditional way: each member pays a set amount, which covers the meal and gratuity and tax.

Occasionally, the club has a guest speaker. At one meeting, Anthony L. Conrad, RCA President and Chief Operating Officer addressed the group. Dr. George H. Brown, formerly Executive Vice President, RCA Patents and Licensing, was the speaker at another.

Above left: George K. Martch, Program Manager, Atmosphere Explorer, explains the design features of one of AED's spacecraft. Above: Hank Hurlburt, Manager, Operations Engineering, Advanced Technology, explains the workings 'Mr Strobl is Editor of TREND anda Consulting Editor to of AED's 28,000 -force -pound vibration test facility to the ex- RCA'ers. the RCA Engineer.

www.americanradiohistory.com The ex- RCA'ers have also held and by 4 p.m. boarded the buses for And mostly it's for fun. It's something meetings at RCA facilities. May is the the return trip. Afterwards, Merrill they like to do and do it just for that traditional month for such visits. For Trainer, "unofficial vice president" of reason. example, in May 1972 and in May the club, said he received many calls 1973, the group traveled to the David commenting on the day's program; "it It is also apparent that with all their Sarnoff Research Center in Princeton was very well received, and the callers activities, they still value their for their meeting. were delighted to have had the oppor- relationship with the corporation and tunity to make the tour; and they were are interested in its various business In May of this year, the club visited the amazed at the size of the facility." endeavors and the people behind Astro- Electronics Division. Three bus them. Through such spontaneous loads of ex- RCA'ers were joined by These meetings of the ex- RCA'ers and voluntary support, this non - others arriving in autos from sur- take but one day a month of their time. organization demonstrates the drive rounding areas. Gathering in the AED What happens to the other days? and spirit that has made RCA ano the lobby, this was the largest turnout yet Well, as we found in our last report, if electronics industry a primary force in the club's three and one -half -year Satan is looking for idle hands, he in this country's development over history - approximately 175 won't find them among the ex- the past half century. members. The following is a short RCA'ers. (Ed. note: For details on the ex- account of their day at RCA's Space RCA'ers club, contact either T.A. Center. A specific example is Wally Poch, Smith, 125 Coulter Ave., Ardmore, formerly at AED and Broadcast Pa., 19003 or M.A. Trainer, 112 First on the agenda, after an informal Systems. In past years, Wally had Colonial Ridge, Moorestown, N.J. welcome, was a trip to the AED made several business trips to Russia 08057.) cafeteria for lunch, where an official and found he "knew enough of the welcome was extended by C.S. (Gus) language so I could do translations." Constantino, Division Vice President Wally continued, "So I'm now doing and General Manager, and Warren translations of technical articles from Wagner, Manager, Industrial Russian to English for an outfit head- Relations. Mr. Constantino then quartered in Washington. I've also provided an overview of activities at done a little of it for the SMPTE. AED and touched on past, present, Apparently there is a need for this and future spacecraft projects at the kind of thing, and as fast as I get these Space Center. done they send me some more."

Ted Smith, "unofficial president" of According to Wally, this type of work issues the club, thanked Mr. Constantino "doesn't pay very much, but in a way it Future and Mr. Wagner for the reception uses all my past experience because accorded to the club by AED and the articles that come in cover a pretty The next issue of the RCA Engineer thanked Wally Poch, formerly with wide spectrum. And it's another way summarizes the recent Corporate AED, for initiating the visit to the of keeping me up to date with what's Engineering Conference at French

Space Center. Ted also introduced going on too. I find I learn things from Lick, Ind., and contains represen- two new, distinguished members of the material they send me. And it's tative papers from several divisions. the ex- RCA'ers: Barton Kreuzer, who rewarding in that respect." Some of the topics to be covered are: headed AED from 1960 -1967, and Dr. George H. Brown. Ted also noted the In discussing retirement with others, Viking communications system presence of Otto Schairer "one of the the response from Tony Maugeri, fiber communications real veterans of the company," who formerly with Broadcast Audio Optical was the senior attendee at 94. Dr. Products, Camden, N.J., was a typical IC's for auto radios Schairer was Vice President in charge one: "I've retired from RCA, but not Pyroelectric detectors of the RCA Patent Department in the from life. It's amazing, I'm busier now Electro-optic laser modulators

30's. than I ever was." Charge -coupled imager Laser recording The retirees then toured AED's Solid state relays facilities. The visitors were treated to And Sam Watson, formerly Manager, views of the ITOS integration area, Corporate Standards Engineering, Discussions of the following themes Atmosphere Explorer integration Cherry Hill, N.J., observed, "One are planned for future issues: area, environmental test facility, com- thing that becomes immediately puter room, manufacturing clean obvious, when you retire, is that there International activities room, and high -resolution lab. are no end of opportunitities available Computer aided design to you on a no -fee basis." Consumer electronics After this look at the sophisticated Many of the ex- RCA'ers are in Automatic testing systems in AED's bailiwick, the ex- business for themselves or are plan- Parts and accessories RCA'ers reassembled in the lobby ning to be in a business of some sort. Automation and process control

www.americanradiohistory.com History and development of the color picture tube E. W. Herold

This paper covers the early history of the shadow -mask color picture tube and its development up to approximately 1973. The shadow -mask tube by then was almost 25 and freedom to use all the technical years old and had established itself as the only successful color display for home expertise that RCA had available. There television. Other color tube types that have been under laboratory investigation, but have was only one catch: Feasibility was to be not achieved commercialization, will not be described; for a more complete discussion demonstrated in three months and a final that includes these types, the reader is referred to the Academic Press book, Color result in six months. It seemed impossible Television Picture Tubes, by Morrell, Law, Ramberg, and Herold.' because, although we had some ideas, none were technologically practicable. But research teams were set up to explore five different approaches that have been described elsewhere.' One available idea had been described some years before in an invention by a German named Flechsig.' He proposed the shadow effect of a grill of wires, with three electron beams, one for each color. It seemed clear to many that Flechsig had never tried to make such a tube, or he would have been embarrassed to put his name on the idea. What appeared to be equally impractical

Reprint RE- 20 -1 -7 Presented at the San Diego National Symposium of SID, May 21, 1974. Final manuscript received February 12, 1974.

Dr. Edward W. Herold, was Director, Technology, Fig. 1 Picture - of one of the first shadow -mask tubes. Research and Engineering at the time of his retirement in November 1972. Dr Herold joined RCA in 1930 in sHADOW -MASK development start- went on at a feverish pace, with many Harrison, N.J., where he engaged in tube researcn. In 1942 he transferred to the newly established RCA ed in September 1949. For anyone who predictions that some day soon a simple Laboratories where he served as Director of the Radio was not involved, it is difficult to ap- and low -cost color display would become Tube Laboratory and, later, the Electronics Research preciate the inadequacy of color displays Laboratory. In 1957, he was placed in charge of the RCA available. To many experts in the group working on Princeton University's C Stellarator at that time. Black- and -white television electron -device field, these predictions project for research on controlled thermonuclear fusion. was well established, with excellent pic- seemed irrational, with no basis in reality. In 1959 he was appointed Vice President, Research, for tures via the cathode -ray tube. For color, Varian Associates, returning to RCA in 1965, where he Yet, six months after starting work, some was a member of the Corporate Engineering for there were Staff the only three ways to display a of these same skeptics (the writer in- past seven years. Dr. Herold received the B.S. in Physics picture. The first was the rotating color cluded) were able to take to Washington in 1930 from the University of Virginia and the M.S. in disc in front of a black- and -white picture Physics in 1942, from the Polytechnic Institute of four color receivers with a half-dozen Brooklyn. That institution honored him with a Doctor of tube. The method was severely limited Science degree in 1961 for his ac- in spare tubes and put on a demonstration "distinguished picture size and required scanning stan- complishments in the fields of electron tubes and solid in which the performance achieved state devices." A Fellow and former Director of the IEEE, dards that were not compatible with startled the entire industry. Color pic- Dr. Herold is a member of the American Physical Society, black -and -white transmissions. A the American Association for the Advancement of second tures were displayed on the shadow -mask method used three Schmidt -optics pro- Science, the Society of Motion Picture and Television color tube, using a compatible -color Engineers, Phi Beta Kappa and Sigma Xi. jection systems with images superim- system essentially the same as the one posed on the screen; this was too ex- now used throughout the world. pensive for a consumer product. Finally, there was the most awkward of all - a combination of three orthogonal The mobilization of corporate cathode -ray tubes viewed in superposi- resources tion through large dichroic mirrors. This arrangement produced a direct -view pic- It all started at RCA Laboratories one ture of about 12-inch diagonal in a day that September, 1949, when Dr. cabinet occupying the volume of two Elmer Engstrom, who then headed the upright pianos! Laboratories, asked the writer to organize and coordinate a company -wide In spite of these inadequate systems, program to develop a color tube. We were work on color television in early 1949 promised unlimited funds, top priority,

www.americanradiohistory.com "ski girl" slide at time of first demonstration (original in color). Fig. 2 - Picture of color receivers at the first demonstration, March 1950. Fig. 3 - Picture taken of was modification suggested by Dr. A.N. (unrelated to Harold Law), who was part competent people are told that the sky's Goldsmith4: thousands of tiny holes in a of the RCA group that worked so hard to the limit - forget money, forget red tape, mask instead of wires, with make the shadow -mask tube a reality. forget lines of authority - their three electron guns spaced 120° apart and However, the three -gun tube had so many enthusiasm has no bounds and even the separate deflection systems corrected for advantages that the one -gun version is no unattainable becomes possible. If the "keystoning ". In our own laboratory, Mr. longer of interest. In March 1950, two shadow -mask idea had not worked out, at A. C. Schroeder` had proposed putting receivers of each type were taken to least two other ideas could have been the three guns so close together that they Washington to demonstrate to the brought to fruition. Unfortunately, there would pass through the same deflection F.C.C." are far too few corporate or government yoke. Even this seemed out of reach executives who have the foresight and the Fig. 2 there are three receivers: A because no one saw how to line up several In nerve to authorize this kind of project. one in the center, the hundred -thousand tiny holes with an black -and -white They usually insist on budgets, cost con- -gun shadow mask at the right, and equal number of phosphor -dot triplets. three trols, cost/ effectiveness analysis and, in the one -gun shadow -mask receiver at the the end, they may spend much more than left. The picture was taken in March 1950 Nevertheless, in our crash program, Dr. would the "crash" program while often at the time of the demonstration. The failing to achieve the desired result. Harold B. Law selected the Schroeder picture quality of the three -gun tube is idea for exploitation. Law was skilled as shown by Fig. 3 taken at the same time. The 1950 tube was, of course, only a start. well as tenacious, and he then made the Compared with today's tubes, the pic- Most of the RCA group had what proved key invention, called the "lighthouse"6 tures were smaller, had only 7 -f I highlight idea. We believed With this device, Law used light to to be an erroneous brightness and required an almost fully there was one way only to achieve low simulate the shadowing of electrons. The darkened room for satisfactory viewing. enough cost to make the color tube light permitted use of photographic and However, the industry reaction was practical, and that was to make every lithographic processes to locate the A typical from overwhelming. comment shadow mask exactly alike, and every phosphor dots in the desired positions the April 1, 1950 TV Digest is quoted as phosphor screen identical with every back of the mask. The lighthouse is still follows: other. We thought that they could then be the basis of today's manufacture of color "Tri -color tube has what it takes: RCA shot mass -produced and any mask would tubes, and it still goes by the same name. the works with its tri -color tube demonstra- work with any phosphor screen. The method worked so well that, in less tion this week, got full reaction it was Although the RCA group was making than the prescribed three months, Law looking for....not only from....FCC...and progress toward that goal, the accuracy had a few square inches of color picture newsmen, but from some 50 patent required in construction of the screen and that proved feasibility, and he had built a licensees...." "So impressed was just about everybody by remarkable performance, that mask made this approach very difficult to first tube with a 7 -in. diagonal screen. In it looks...as if RCA deliberately restrained more months, a few hundred other carry out. three its pre- demonstration enthusiasm to gain people, working seven days a week, had full impact." hélped produce a dozen or so tubes with a Another approach to mass production, I2 -in. diagonal picture of remarkable This early experience deserves attention which abandoned the interchangeability quality7' ". A picture of one of these first both because it was so dramatic, and feature, was the result of the ingenuity of tubes is shown in Fig. 1. It used a 16 -in. because there's a lesson for all of us. This Mr. Norman Fylerl', in late 1953, at CBS metal bulb and had a flat color screen and project was an example of how an ap- Hytron. Instead of the screen being shadow mask mounted inside. parently unattainable result can be placed on an internal flat glass plate as in achieved by use of unlimited effort. In the RCA approach, Fyler and his A one -gun variation of the shadow -mask fact, of course, there's no such thing as associates built a shadow -mask tube with tube 10 was devised by Dr. Russell R. Law unlimited effort; yet, when a group of a curved mask and direct photo-

www.americanradiohistory.com deposition of phosphors on the curved vances were incorporated in the first The Trinitron inside face of the tube. A crucial aspect of large -scale production type introduced the system was that the mask was destined by RCA in 1954.'4 The tube had 70° About 1968, as a result of work at the for use with that screen only. Exact deflection, with a 21 -in.- diameter round Sony Corporation in Japan, a different duplication of parts was not necessary metal bulb and the phosphors were design of shadow -mask tube was in- and the picture was larger more and photo- deposited on the curved glass troduced under the name Trinitron. "'18 appealing on the face of the tube than on faceplate. This type was used in most of As with many other color tube the internal flat -screen plate in use by the the early color sets that established the developments, the Trinitron was also the RCA group. Fyler's design is now the NTSC color system as a commercial result of a team effort, in this case by S. standard way of making color tubes. It is success. A few years later, 1958, the metal Miyaoka, A. Ohkoshi and S. Yoshida. In interesting that, although Harold Law envelope was replaced by a glass one,'6 4, left, is shown the made his first screen by a settling process, Fig. at the traditional made possible by the development of a he had also invented the direct photo - round -hole shadow -mask arrangement. new frit -glass seal by Corning Glass: This deposition of phosphors" and mentions It uses three essentially separate electron a it in his original paper.6 frit fuses at low temperature to make the guns in a delta configuration pointed at seal. Upon further heating it converts to a the center of the screen. Shown crystalline material which provides a schematically at the right in Fig. 4 is the RCA technological advances ceramic -like bond that cannot again be Trinitron gun which uses a single large - parted. As with some of the other early diameter lens with three in -line electron Although not the first with the curved developments, the frit seal remains to the beams from the one gun. The shadow mask, the RCA group had its own set of present day. mask now consists of vertical strips, and successes in the early 1950 -1954 period. the phosphors are deposited in vertical Several of their developments must be Other important developments that are lines. mentioned because they have remained in only touched upon here were the slurry use ever since; without them there is some process that mixes the phosphors with the The Trinitron has three advantages over question whether the shadow -mask photoresist and the correction lens that is the traditional design. The large lens principle would have reached its present used in the lighthouse to preserve the diameter reduces aberrations, so that the predominance. The first has to do with match of phosphor dots with electron electron spot is more easily kept small. the convergence or coincidence of the beam landing positions, particularly un- Secondly, the vertical apertures in the three beams. Obviously, unless the red, der large -angle beam deflection con- mask have no horizontal cross ties or green, and blue pictures coincide, i.e., ditions. But beyond even these items, the structure, so that the vertical resolution is converge over the entire screen, a poor period of 1954 to 1968 led to almost totally unaffected. A third advantage is picture would result. The solution in- unbelievable advances in manufacturing that convergence of three in -line beams is volved pointing the three guns to the technology; costs were reduced, quality simpler than with the delta -gun arrange- center of the screen and providing both and performance were greatly increased, ment. Convergence of in -line beams is static and dynamic adjustments, via static and picture size, deflection angle, and discussed in more detail later. Pictures and alternating magnetic fields. The im- brightness became comparable to those produced by the Trinitron are reputed to portant invention here was the incorpora- of black- and -white tubes. There were be excellent. tion in the three electron guns of internal hundreds of other detailed improvements pole pieces that permitted each beam to in guns, compensation for thermal ex- Along with the advantages, there are be independently adjusted.14 The earliest pansion of the mask, phosphor deposi- some shortcomings. Because the mask shadow -mask tubes were easier to con- tion and mask making. Major advances has continuous vertical apertures, it can- verge because they used only a 45° were made in color phosphors, notably not be spherically curved and be self - deflection, but this made by use of rare -earth elements in the red the tubes supporting, as with the round -hole mask. awkwardly long. By devising - phosphor. The shape of the picture tube the internal Thus, it is curved cylindrically and face was made rectangular, and the tubes pole -piece gun, it became possible to go stretched on a relatively heavy frame. to 70° deflection and ultimately to 90° became much shorter by increasing the Trinitron designs are often longer than and more. deflection angle to 90° . From a slow start, traditional ones of the same screen when few programs were broadcast in diagonal and deflection angle. In sum- A second RCA development that remains color, receiver sales and color broadcasts mary, the good technical performance is in use to this day relates to the curved progressively increased, and many accompanied by a higher cost of mask principle employing photographic manufacturers entered the field of color manufacture and a less flexible receiver deposition of the three color phosphors; tube manufacture. By 1967, about 15- configuration. such deposition requires three separate million color tubes had already been exposures. The shadow -mask must be produced and the shadow -mask tube was mounted and removed again for each finally established as a commercial as well The matrix screen exposure, without any shift in alignment as a technical success. when it is remounted. This is an exacting It has long been recognized that the requirement, but a solution was found by In the most recent period, from 1968 to "whiteness" of phosphor materials is a mounting tapered metal studs at the sides 1973, design changes continued but there disadvantage in that it diffuses back any of the faceplate cap.15 The studs are were four dramatic developments that ambient light and reduces contrast. engaged by circumferential springs on the deserve more detailed scrutiny. These Transparent phosphors were tried in an lightweight frame that holds the curved four are the Trinitron, the black- matrix attempt to reduce the back- scattered light shadow mask. phosphor screen, 110° deflection, and the but were far too inefficient. For a long self- converging Precision In -Line system. time, the only low -cost solution was the These two important technological ad- These are taken up in turn. use of a light- absorbing glass faceplate -

www.americanradiohistory.com BLUE ELECTRON GUNS ORIGINAL SYSTEM MATRIX SCREEN SYSTEM

BEAM PERMITTED BEAM PERMITTED BY MASK OPENING BY MASK OPENING ''GREEN RED DIRECTION OF T--17 1 RED . SCANNING LINES BLUE.,, ",GREEN 1I r- r1 METAL ' MASK -y \ I h/ 1 o

TANGENT PHOSPHOR BLACK MATRIX DOTS

IF BLACK MATRIX COVERS ( GIR)'{ R¡B B1 PHOSPHOR 1/2 OF AREA: JR. LIGHT, PHOSPHOR DOTS

B I G.R/0B)R REDUCED BY T Í6)IG ON 6LASSRS BRIGHTNESS = B0 G PLATE G B B , FACE G B R BÌG,..E-',)J ÌR G Ì B R FOR T= CONTRAST =2X DIFFUSED 2 (.,)aOr-' AMBIENT, BRIGHTNESS = Ete 1 REDUCED BY T2 J D CONTRAST = 2X

I i N ) . t "GREY" GLASS. DELTA GUN IN -LINE GUN TRINITRON TRANSMISSION =1'

Fig. 4 - Trinitron system, compared with conventional delta -gun system. Fig. 5 - Grey -glass action and comparison with matrix screen.

the so- called "grey" glass. Grey glass round openings defining only the central matrix methods to achieve what he works because the ambient light passes useful area of the phosphor dots. This believes to be a best compromise. The through the grey glass twice, but the means that the maximum possible area of matrix screen made a major advance in phosphor- emitted light goes through black is provided. However, to provide picture -tube performance and is now in only once. In Fig. 5 it is shown that, if the tolerance for possible beam position inac- extensive use in the United States. In faceplate has a transmission T of the curacy, it is now necessary to make the Europe, matrix screens have not yet order of 0.5, the phosphor light output is beam areas larger than the matrix taken over. One reason may be that, with cut in half; the diffused ambient goes openings, and this is done by opening up their 50 -Hz standards, high brightness down to T2, i.e., to one -quarter. Thus, by the mask holes to be larger than the holes makes flicker more noticeable. losing half the brightness, the contrast is in the traditional non- matrix mask. In the doubled. Up to 1969, this was the only first RCA matrix tube, the beam spots, method in use to enhance contrast in and therefore the mask holes, are smaller color tubes. than the phosphor dots as defined by the Large -angle deflection matrix; and the system is said to have Fig. 5 also shows one of the hundreds of "positive" tolerance. With the Fiore - One of the disadvantages of the standard thousands of phosphor-element triplets Kaplan matrix, the mask holes are larger cathode -ray tube when designed into a as they are located on the faceplate. In than the useful size of the phosphor dots display console has always been its front - practical tubes, the mask openings must and the system is said to have "negative" to -back length. It has been desirable to be made smaller than the phosphor dots tolerance. Tubes with this system are increase the deflection angle as much as so that slight inaccuracy of position will often loosely called "negative matrix" possible to reduce this length. However, a not change the color. The tolerance types. large angle makes convergence of the required can be as much as 25% of the three beams more difficult and requires diameter, so that the useful part of the The manufacturing procedure for the high deflection power unless the neck phosphor covers only about 50% of the negative -tolerance matrix screen is con- diameter is reduced in inverse propor- screen area. Color -tube workers had long siderably more complex than for the tion, which produces a new set of recognized that it would be desirable to traditional non -matrix type, or for the problems. Fortunately, there is also a use black on any portion of the screen not positive tolerance type, but the result is performance advantage in a wider deflec- needed for phosphor. This was first most gratifying. Because the screen now tion angle in that the beam has a shorter applied on a different type of color tube, has about 50% of its area covered by throw and the spot remains small for a called the line- screen index tube, black, the back -scattered ambient light is higher current, so that the picture can be developed by Philco but never put into cut in half just as with 50% grey glass, but made brighter. Conversely, the picture production. About 1969, both RCA and there is now no reduction in brightness. can be sharper for the same brightness Zenith announced shadow -mask tubes In other words, the picture is twice as when the deflection angle is increased. that used a black matrix on the faceplate, bright as the grey -glass tube for the same The advance from the original 45° angle with round openings that permitted the contrast. tube to 70° and then to 90° has been phosphor dots to be seen. Only the Zenith mentioned. In no way should achieve- version, published in 1969 by J. P. Fiore The numbers used are illustrative only. In ment of these angle increases be con- and S. H. Kaplan19 is described, because practice, there are many tradeoffs sidered easy; they were all difficult. But it gives the largest improvement and is between the permissible tolerances, the the goal remained even higher, and it was now available from most manufacturers, matrix open area, and the glass faceplate first achieved in 1970 by an RCA tube of including RCA. transmission, which result in tradeoffs 18 -inch diagonal and 110° deflection20. between brightness and contrast. Usually The same year, a 25 -inch tube with 110° The matrix used by Fiore and Kaplan had the tube designer uses both grey glass and deflection was announced by Philips21

www.americanradiohistory.com ELECTRON BLUE RED GREEN and one year later in a different form, by GUNS RCA` ''. The Sony engineers went just a I little further, to a 114° Trinitron''. In all e these cases, the deflection problem was DIRECTION OF aided by new yoke designs and by new SCANNING LINES solid state devices for horizontal deflec- The reduced cabinet depth METAL tion. MASK permitted by these large -angle systems is considered highly desirable and, fortunately, the picture is improved as well.

The Precision In -Line system

The most recent advance in color tubes PHOSPHORS ON GLASS BI covered in this paper is known as the FACE PLATE N Precision In -Line system, announced in I ) 1972 by RCA.24 This system is also a radical departure from the traditional shadow -mask design. As with the Trinitron, the beams are in a horizontal GUNS AND SCREEN line, as shown in Fig. 6, and the phosphors are in the form of vertical Fig. 6 - Precision in -line system. stripes. The mask, however, in distinction to that of the Trinitron, uses slits with horizontal supporting ties or webs. These webs are sufficient to provide overall rigidity to the mask so that it can be self- supporting and can be curved spherically. Thus, the manufacturing technology retains the low cost of the round -hole mask. In addition, the system solved one of the major problems of shadow -mask tubes, namely the convergence of the three beams over the entire screen.

With the traditional delta gun and round - hole mask, it is necessary to add special dynamic convergence hardware and ap- GJN PLANE DEFLECTION PLANE SCREEN propriate adjustable ac signals in order to converge the three beams. The adjustments are made by the receiver ACTION OF ASTIGMATIC TOROIDAL manufacturer for each individual tube YOKE DEFLECTION and sometimes must be touched up in the home. Replacement of the picture tube Fig. 7 - Astigmatic yoke action. requires complete readjustment. In the typical traditional system, there are 12 such adjustments needed to obtain red,

0 green, and blue pictures that are properly superimposed all over the screen. It is / remarkable that a skilled technician, using a video dot pattern and a cross -hatch EID ITN 30 / pattern, can get good convergence in a 20 matter of minutes. Even so, convergence \-' has always been an annoying difference between the simple black- and -white picture tube and the three -gun shadow -mask LIGHT OUTPUT OF SHADOW -MASK COLOR PICTURE TUBES tube.

Use of an in -line gun, as has already been mentioned, simplifies convergence somewhat, but the ultimate in simplifica-

I tion came about by designing a special 1950 1955 1960 1965 1970 975 -line YEAR deflection yoke to complement the in gun and the line- screen system. When the Fig. 8 - Plot of brightness increase from 1950 to 1973. yoke is aligned with the gun in the

www.americanradiohistory.com Precision -In -Line system, no dynamic is the screen brightness. In Fig. 8, the technical and management contributors convergence adjustment is needed and average brightness in a picture is plotted to this development. As the history of the none is provided. The yoke is aligned on against the calendar year. The first tubes past two decades is being studied, the tube neck and can be permanently of 1950 had about 2 fl, average, and 7 fl in perhaps others will record their ex- affixed in the tube factory. The combined the highlights. By 1973, 100 fl average was periences so that the whole story will tube and yoke can then be plugged into a obtainable, with several times as much in become known. receiver with almost the same simplicity highlights. The dotted line on the ex- as with a black -and -white tube. The ponential plot shows that the brightness principle employed is shown in Fig. 7. has been increasing at a rate of 17% per The yoke is purposely made astigmatic. year. It is small wonder that other color - References An astigmatic system, as we know, tube concepts have failed to compete in produces an image that is distorted from the commercial market, even though they I. Morrell. A.M.; Law. H.B.: Ramberg. E.G.; Herold, E.W.: Color Television Picture Tubes (Academic Press: N.Y.; that of the object. In this instance, a are technically feasible and much 1974) circular object is distorted into a vertical laboratory work has been done on them. 2. See a series of I I papers in Pror. IRE. Vol. 39 (Oct. 1951) line. If the three beams lie on a horizontal pp. 1177 -1263; also in RCA Review. Vol. 12 (Sept. 19511 pp. 445 -644. diameter of the object circle, they then 3. Flechsig. W.; "Cathode Ray Tube for the Production of Multicolored Pictures on a Luminescent Screen," German converge to the center of the vertical line. 1 In Table the milestones in the history of Patent 736,575 (filed 1938); see also the corresponding The only trick is to assure accuracy in the the shadow -mask tube are listed. What French Patent 866.065 (issued 1941). yoke field pattern so that the astigmatism appeared as an impossibility in 1949 has 4. Goldsmith, A. N.: U.S. Patent 2.431, 115 (filed 1944. issued occurs uniformly over the entire screen. I947). evolved as one of the most remarkable 5. Schroeder. A. C.: "Picture Reproducing Apparatus.' U.S. Precision yoke fields were obtained by a and successful developments ever made Patent 2, 595, 548 (filed 1947. issued 1952). ferrite core with slotted end pieces and in the history of the electron tube. It is 6. Law. H.B; "A three -gun shadow -mask color kinescope." Pror. IRE, Vol_ 39 (Oct. 1951) pp. 1186 -1194. See also toroidal windings in which every turn is also interesting that two of the major Law. H.B.; "Art of Making Color Kinescopes," U.S. exactly positioned in its correct slot. With developments in recent years, the in- Patent 2. 625. 734 (filed 1950, issued 1953). this the yokes are all es- 7. Freedman, N. 5. and McLaughlin. K. M.; "Phosphor - construction, tegrated circuit and the shadow -mask Screen Application in Color Kinescopes:' Prue. IRE, Vol. sentially identical. In the tube factory, the tube, have both come about through 39 (Oct. 1951) pp. 1230 -1236. yoke is positioned and then permanently of 8. Moody. H.C. and Van Ormer. D. D.: "Three -Beam Guns applications photolithography. Those for Color Kinescopes :' Pro, IRE, Vol. 39 (Oct. 1951) pp. attached to the tube neck. Center -screen of us in the field of science and engineer- 1236 -1240. static convergence is built into the gun ing 9. Barnes, B. E. and Faulkner. R. D.; "Mechanical Design of tend to credit ourselves with the Aperture -Mask Tri -Color Kinescopes," Pror. IRE, Vol. and is touched up with small external brains, the ingenuity, and the foresight to 39 (Oct. 1951) pp. 1241 -1245. magnets, also in the tube factory. propose and carry out our new I(I. Law. R. R.; "A One -Gun Shadow -Mask Color Kinescope." Pror. IRE, Vol. 39 (Oct. 1951) pp. 1186 -1194. Thereafter, the tube is ready for shipment developments. In the case of the shadow - I I. "General Description of Receivers which Employ Direct - and has eliminated the 12 dynamic con- mask tube, much credit must go to the View Tri -Color Kinescopes" RCA Review, Vol. 11 (June 1950) pp. 228 -232. vergence adjustments required by the executives and financial entrepreneurs 12. Eyler, N. F.; Rowe. W. E.: and Cain, C. W.; "The CBS traditional shadow -mask tube. who were willing to take the risk in Colortron" Prof.. IRE. Vol. 42 (Jan. 1954) pp. 326 -334. supporting the tremendous investment in 13. Law, H. B.: "Photographic Methods of Making Electron - Sensitive Mosaic Screens.: U.S. Patent 3, 406. 068 (filed The Precision In -Line system can be manufacturing technology that was re- 1951. issued 1968). thought of as an example of integration, quired. Without such an investment, 14.Seelen, H. R.: Moody. H. C.; Van Ormer. D. D.: and Morrell, A. M.; "Development of a 21 -inch Metal - not as complex as an integrated circuit, color television as we know it today Envelope Color Kinescope :' RCA Review. Vol. 16 (Mar. but sharing the advantage of a factory - would not have been possible. The two 19551 pp. 122 -139. 15. Janes, R. B.: Headrick. L. B.: and Evans, J.; "Recent designed system that greatly simplifies men to whom the greatest credit in this Improvements in the 21AXP22 Color Kinescope." RCA color receiver production and respect must be given are David Sarnoff Review. Vol. 17 (June 1956) pp. 143 -167. In we see a and Elmer Engstrom, the two RCA 16. Smith. C. P.; Morrell, A. M.; and Denny. R. C.: "Design maintenance. this system, and Development of the 2ICYP22. 21 -inch Glass Color portent of the future, in which color tubes executives who steadily and consistently Picture Tube" RCA Review. Vol. 19,(Sept. 1958) pp. 334- are no more difficult to use and install pushed ahead on the shadow -mask tube 348. 17. Yoshida. S. and Ohkoshi, A.: "The Trinitron -A New than is the ordinary single -beam cathode - at a time when it looked as though it Color Tube" IEEE Trans. on Broadcast and TV ray tube. would never become profitable. Receivers. Vol. BTR -14, No. 2 (July 1968( pp. 19 -27. 18. Miyaoka, S.. Ohkoshi, A.: and Yoshida, S.; "The Trinitron A New Color TV Tube." IEEE Student Jour., Vol. 8

I May 1970) pp. II -15. 19. Fiore. J. P. and Kaplan- S. H.: "The Second- Generation Conclusion In this brief space, it is impossible to Color Tube Providing More Than Twice The Brightness adequately acknowledge the many other and Improved Contrast," IEEE Trans. on Broadcast and The progress that has been made in the 24 TV Receivers. Vol. BTR -15, No. 3 (Oct. 1969) pp. 267 -275. 20. Masterton, W. D. and Barbin. R. L.; "Development and years since 1950 is remarkable. Cost and Performance of the RCA 19V and 18V 110° Color Picture picture quality have been tremendously .tubes and Deflection Yoke.' Electronics, Vol. 44, No. 9 (Apr. 26, 1971) pp. 60-64.

I in the improved. A large range of picture sizes, Table - List of milestones development of the 21. Philips Product Information. No. 13; "I10° ColourTelevi- shadow -mask tube. from 25 -in. diagonal on down, were made sion. Picture Tube and Deflection Principle.' N. V. Philips Gloeilampenfabrieken. Electronic Components and has Materials Division. Eindhoven, The Netherlands. available, and the length of tubes 1950 - First tube demonstrated 22. Thierfelder, C. W.; "RCA Large- Screen, Narrow -Neck been greatly reduced. Each year, ap- 1953 - Curved mask with phosphor on faceplate 110° Color Television System." IEEE Trans. on Broadcast proximately 20 million tubes are being 1954 21 -inch, 70° deflection tube and TV Receivers, Vol. BTR -17. No. 3 (Aug. 1971) pp. 141- manufactured, worldwide, representing a 19M-1963 Period of manufacturing advances 147. 23. Yoshida. S.: Ohkoshi. A.; and Miyaoka. S.; "A Wide - 1964 90° Deflection. rectangular tube $2- billion part of the consumer elec- Deflection Angle (114°) Trinitron Color- Picture Tube." tronics industry. 1965 Rare -earth phosphors IEEE Trans, on Broadcast and TV Receivers, Vol. BTR -19 1968 - Trinitron (3 -beam gun. striped phosphor) ( Nov. 1973) pp. 231 -238. 1969 Black matrix 24. Harbin, R. L. and Hughes. R. H.; "A New Color Picture Tube System for Portable TV Receivers." IEEE Trans. on Perhaps the simplest factor for showing 1970 110° deflection - Broadcast and TV Receivers, Vol. BTR -18, No. 3 (Aug. the improvement in picture performance 1972 Precision in -line system 1972) pp. 193 -200.

www.americanradiohistory.com High-voltage srlir dvi ces open doors to eléttronic ignition M.S. Reutterl J.S. Vara

A key element in the success of the electronic ignition system has been the availability of highly reliable, low -cost, high -voltage silicon power transistors in quantities commen- surate with automotive production line requirements. Automotive designers have used these devices to produce systems that perform effectively and that are more reliable than electromechanical systems with point contacts.

CREDIT for the introduction to the included, approximately 10 million original- equipment market of an all - original equipment ignition systems will electronic ignition system must go to have become solid state, for an 85% Chrysler. The success of their optional penetration by solid -state devices into the all- silicon transistor system introduced in market in only a little over three years. 1972 led to the use of a similar system as This is a truly remarkable response, and a standard equipment on all of their model validation of the success and benefits of 1973 cars. This bold step by Chrysler has the electronic ignition system. led to the "all- out "effort by all other auto makers to incorporate the electronic ignition on a "no- delay" basis on their models Evolution of ignition systems as well. The ignition system of an internal com- The beneficial effects of long-term timing bustion engine must fulfill two basic accuracy coupled with the aid to emission functions: 1) Release energy in the form control provided by higher firing voltages of a spark and 2) Determine when that are the prime technical incentives behind spark is to occur. In conventional adoption of the electronic ignition. By breaker -point ignition, Fig. 1, the points model -year 1975, what began as one car perform the switching function of in- maker's revolution will have evolved into

a standard feature offered by all four Reprint RE- 20 -1 -16 domestic car makers. When trucks are Final manuscript received May 24, 1974.

www.americanradiohistory.com SWITCH

HIGH VOLTAGE BATTERY TO DISTRIBUTOR + BATTERY BALLAST RESISTOR HIGH VOLTAGE IGNITION TO SPARK COIL PLUGS VIA BREAKER POINTS DISTRIBUTOR IGNITION COIL

POWER DISTRIBUTOR PICK - UP SIGNAL TRANSISTOR SHAFT ELEMENT CONDITIONER CAPACITOR SWITCH

Fig. 2 - Block diagram of electronic ignition system. ENGINE DRIVEN CAM

Fig. 1 - Conventional cam -point ingition system.

terrupting the coil current, thereby magnet forms a closed magnetic path life are very close. To overcome this producing the high -voltage spark. The through the coil pole piece and reluctor disadvantage a trigger circuit is required cam on the distributor shah provides the wheel. As the distributor rotates, that is capable of accurate triggering over timing function. This is the lowest cost magnetic coupling varies, and a signal is a wide range of signal levels and that has a system, hut it has several mayor disadvan- generated in step with engine rotation. A high degree of noise immunity to prevent tages. inherent in all contact -triggered problem encountered with this type of triggering at low noise levels. ignitions is timing error resulting from pick -up results from sensitivity of the contact erosion and rubbing -block and output signal to speed. The transducer Photoelectric pick -up distributor -shaft wear. Frequent, costly must produce a usable output signal at tune -ups are needed to maintain engine distributor speeds as low as the 15 r, min Another form of triggering used to a performance and compliance with encountered during cranking. To achieve lesser extent in the after -market employs present -day emission standards. the required performance economically, a light- emitting diode as shown in Fig. 4. the air gap between rotor and stator must A parallel beam of radiation passes To overcome the disadvantages of the be very small. Thus, the tolerances to between detector and source; triggering is conventional cam -point system, several which the distributor assembly must he accomplished by means of a rotor which contactless systems have been developed. built and maintained throughout service interrupts the beam. By arranging the The pick -up elements of such systems are required to generate a signal that varies in

step with the mechanical rotation of the John S. Vara, Power Transistor Applications Engineer- Maurice S. Reutter, Power Transistors, Vehicular Market distributor drive shaft. Contactless ing. Solid State Division, Somerville, NJ, received the Planning, Solid State Division, Somerville, NJ, attended systems require additional electronics BSEE in 1965 from Fairleigh Dickinson University. He Iowa State College, Ames Iowa. He joined the RCA worked for Tung -Sol Electric in Rectifier and Transistor Service Company as a Radar Engineer in 1942. His between the pick -up elements and the Applications prior to joining RCA in 1966 as an background in commercial communications with RCA power -transistor switch to define a Applications Engineer for Power Transistors. He is covers transmitter design. field sales and product plan- currently Project Leader in the Power Transistor ning for the RCA line of mobile communications ecuip- precise trigger point and to amplify the Applications Automotive Group. Mr. Vara received the ment. This includes promotion and marketing of RCA's small signals picked up- hence, elec- RCA Electronic Components and Devices 1968 first transistorized portable receiver "The Per - Engineering Team Achievement Award for Power ". In 1959 he joined the Solid State Division in a tronic ignition. Fig. 2 shows a simplified sonalfone Transistor Engineering. field sales caoacity in Detroit, Michigan. Since that time block diagram of an electronic ignition Mr. Reutter has witnessed and participated in the revolution that brought about the diversity of electronic system. Authors Vara (left) and Reutter. applications of solid -state devices to the automobile.

Contactless systems

Magnetic pick -up

The most common form of contactless ignition system employs a variable reluc- tance transducer, Fig. 3. The pickup unit

PICK -UP COIL

AIR GAP

RE LUCTOR WHEEL

Fig. 3 - Magnetic pick-up.

www.americanradiohistory.com ENGINE DRIVEN ROTOR

CONTROLLED automotive manufacturer had to con- CURRENT i_ SOURCE ENGINE sider performance in terms of lower DRIVEN ROTOR exhaust emissions, better cold weather o OSCILLATOR starts, and lower maintenance. While capacitive- discharge systems were,

LIGHT TUNED EMITTING perhaps, more desirable from a semicon- DETECTOR CIRCUIT DIODE ductor viewpoint (lower stress levels), (LED) Fig. 5 - Conductance detector. ignition performance was unacceptable because of the typically short spark dura- TO tion inherent in these systems. Short BATTERY DISTRIBUTOR spark durations can cause engine misfire IGNITION Fig. 4 - Photoelectric pick -up. with present -day air -fuel mixtures. (With COIL the move toward leaner ratios., the 2N63B5 beam parallel to the distributor shaft, the capacitive- discharge system was just npt effects of axial rotor movement on timing acceptable; therefore, the remainder of RCA accuracy can be eliminated. Also, radial this article will consider only inductive DEv NO TA8847 movement of the rotor will have no affect ignition systems.) provided the rotor disc continues to cut Two types of inductive ignition circuits the beam completely. With present radia- *MONOLITHIC DARLINGTON are shown in Figs. 6 and 7. Both designs tion sources, the beam width may be (Al use a zener clamp from the -to- several millimeters; this dimension collector

base to protect the output transistor from 15 represents a substantial shaft angle. voltage, and both are now in use Special circuit techniques are required to excessive by car manufacturers. Current in the select a fixed trigger point within the regulated type, Fig. 7, is limited to ap- -W+á beam. óF 5 proximately amperes. The power I ° transistor in this circuit is required to ri Conductance detector handle considerably less energy under worst -case conditions than its counter-

A third contactless system, shown in Fig. part in the saturated switch circuit. The IOO 200 300 400 VCE - VOLTS 5, is the conductance detector. This trade-off is cold- weather starting perfor-

system works on the principle of loading mance and higher dissipation in the lb) the Q of a tank circuit of a tuned transistor. Fig. 6 - (a) Saturated -switch inductive- ignition circuit oscillator. The system advantages are (b) worst -case load line (open secondary, ballast resistor shorted). noise immunity and a signal excellent The hostile whose amplitude is independent of speed. environment TO DISTRIBUTOR The ambient temperature under the hood Dwell of an automobile varies widely. A low temperature of -30 °C ( -22 °F) can be Dwell is the period of time during which expected in a parked vehicle. The battery current flows in the ignition coil ambient temperature rise after start -up before each spark, thereby providing a will be slow at this temperature extreme; reference for the measurement of stored therefore, the low- temperature design energy available for spark voltage. A goal which has been adopted is -30 °C. system with cam and breaker points High -temperature extremes depend on generates a fixed percent dwell indepen- several factors, including ambient con- performance. dent of r/ min. The dwell is a compromise ditions and cooling- system between excessive current at idle and too Components reasonably isolated from little current at high r/ min. Mos' of the the exhaust manifold can be expected to pick -ups which replace the ignition points reach 125°C (257 °F). Mechanical shock, (a) do not provide appropriate dwell signals; vibration, and exposure to chemicals are the dwell must be generated electrically. other considerations; however, a One approach is to make the electronics category deserving special attention is the duplicate what the points did; another electrical environment. From an more simplified approach is to produce a electronic- system viewpoint, the must constant off time. automotive electrical environment be rated as severe; Table I summarizes some of the reasons. Circuit selection capacitive - Output- transistor re- 100 200 300 400 vs inductive quirements VCE - VOLTS (b)

Fig. 7 (a) Current -regulated -switch inductive -Ignition I n sclectirig an ignition circuit, the Operating voltages in the automobile can - system: (b) worst -case load line (open secondary. ballast resistor shorted).

www.americanradiohistory.com vary from 6 V during cold -weather manufacturer has developed elaborate Stress, combined -environment, and starting to 16 V during normal running. qualification and production- validation reliability-sequence tests are applied to The worst -case high -line voltage can procedures to achieve reliability goals of produce failures so that design margins climb to 24 V for short periods when two 0.15% replacement in warranty. Obvious- can be determined. Once design goals batteries in series are used for booster ly, design goals must be better than this. have been achieved, real -time indicators starts. The effects of this wide range of are established in production to monitor battery voltage is seen in both ignition The components in the automobile's igni- capability on a lot -by -lot basis. The type performance and power- transistor tion system that are subjected to the of indicator used to monitor all ignition specifications. During cold-weather highest stress levels are the power power transistors is thermal -cycling, an starts, when high energy is needed, the transistors. Thus, power transistors for accelerated intermittent life test designed battery voltage is low; thus, a low voltage automotive use must undergo a reliabili- to produce information in three days for drop is needed at the power stage in series ty test program that reflects the realistic use in process control. with the ignition coil to get the required demands of the environment in which the coil current. At the same time, base drive device is to be used. Table III outlines the to the transistor is down and, to com- type of test program designed by Solid High -voltage ignition switch pound the problem, the forward base - State Division's Reliability Engineers emitter voltage drop is high. Good cold - and used to evaluate power transistors The device developed by RCA for weather starts require high transistor used in ignition systems. The intent of the ignition -coil current switching is a current at low battery voltage. When a reliability evaluation outlined is to test multiple -epitaxial pi -nu (n -v) power 24 -V boost is applied, the transistor device design at maximum levels. transistor. The transistor differs from current will rise proportionately. In a saturated- switch ignition, for example, a 5 -A start requirement can result in 15 -A peak currents. Table I - Characteristics of an automotive electrical Table II - Typical specifications for a saturated- switch- environment. ignition output transistor. The output transistor must be capable of handling high voltage and high current, Peak collector current 15A simultaneously, during worst -case load Automotive line -voltage characteristics condition (open spark plug). Typical Condition Voltage Peak base current 5A requirements for the output stage of a 400V ,vormal running 16V max v rea saturated -switch ignition are shown in 14.411 nomial Table II. 10V min p'ean 5V max Cold cranking 6V '', at,e,l @ / =5A. /v 850mA 1.3V @ Ty = -30°C. +25 °C -20 °F ( -29 °C) Performance and l'ac @ Ir = 5A, Vrc = 1.3V 1.5V max Rooster starts 24V. 3 min. reliability objectives nr Reverse polarity 1, ta @ Vet' = 390V. 15 =300 ImA Based on cost alone, a from change the Functional test conventional cam -point ignition to elec- Automotive voltage transients tronic ignition is not justified. However, emission standards and service Condition Max Time restrictions imposed on the automotive amplitude constant industry have made electronics more Load dump 125V 45 ms attractive. Thus, in developing an elec- 100% functional test for I second in a simulated worst -case Alternator -75V 30 ms ignition circuit: i.e. 18- to 22 -V battery with coil secondary tronic ignition system, the stated goals of field decay upen and ballast resistor shorted. the automotive manufacturer are performance and reliability. For the consumer, the added cost of the solid -state ignition Table Ill - Outline of reliability evaluations performed on ignition power transistors. buys lower maintenance costs and better engine performance. Claims of up to 35% I SEQUENCE more cranking voltage for cold -weather MECHANICAL ENVIRONMENTAL LIFE NEW TESTS AND STRESS COMBINATIONS TESTING starting are common. (In cold weather, when cranking speed is low, points in the I Centrifuge Relative Hum. Storage Temperature cycling. moisture plus conventional system open slowly and Impact Shack Moist. Resist. Reverse Bias Operating switching Vibration Temp. Cycling Operating Lilt Thermal fatigue significantly degrade the output voltage Lead Pull Acc. Operating Lite of the ignition -coil secondary.) Solder bility

The automotive manufacturer is concerned about reliability for three reasons: warrantly STEP STRESS AND /OR CONSTANT OVERSTRESS TESTS costs, safety, and image; all WHENEVER APPROPRIATE specifications and validation testing focus on these important areas. To meet DATA ANALYSIS the reliability requirement, the car FAILURE ANALYSIS

www.americanradiohistory.com other designs in that the depletion regio!. sion. Some of the new efforts in design, for variability of timing by electronic which supports the electric field extends such as integral mounting of the elec- means. These controls will be an integral into both the n -type v- collector region tronics in the distribution housing, are part of overall engine controls for emis- and p -type 7r-base region, thereby leading to reduced system costs; other sion and fuel- management systems. The providing increased volt- ampere efforts, such as reducing component continued demand for improved reliabili- capability at a given current rating. count, will lead to both cost and ty will remain the paramount goal in all Higher voltage ratings are possible reliability advantages. future developments. because both the base and collector regions are used (instead of collector The use of a high -voltage monolithic only) to support the applied voltage. Darlington as the output switch is both Good current References handling results from the logical and cost effective. It means the use lower collector resistivity needed for of one integrated device instead of two or equivalent voltage ratings. Figs. 8 and 9 I. Gallace, L.J. and Vara, J.L; "Evaluating Reliability of three separate components, thus saving in Plastic - Packaged Power Transistors in Consumer show device construction and compare space and mounting costs while enhanc- Applications ", RCA Technical Publication ST -6184 (June, electric -field distribution in conventional 1973). ing reliability. The consolidation of all of 2. Denning, R. and Moe, D.A.; "Epitaxial (n -v) NPN High - and 7r -1, epitaxial structures. the signal -processing portions of the Voltage Power Transistors". RCA Technical Publication ST-4287. system into one integrated- circuit For maximum power handling capability 3. Carlson, D.W. and Doelp, Jr., W.L.; "A Successful package will yield similar advantages. Electronic Ignition System Thru Fundamental Problem and long life, the device pellet is mounted Analysis". Society of Automotive Engineers, Paper No. 140154 (February- March, 1974). on a copper heat spreader in a steel TO -3 4. Moore, J.H. and Longstaff- Tyrell, J.; "Engine Position hermetic package by means of a nickel - The integration of the entire system into Tranducers for Electronic Ignition." Society of Automotive lead metallurgical system. The clips used the distributor housing, one of the Engineers, Paper No. 740153 (February - March 1974). it for the emitter -base contacts are joined to greatest challenges of the future, while 5. Myers, R.S.; Electronic Ignition Comes of Age ", RCA the contact areas with a high lead -content provides many logistic and cost advan- Reprint RE- I8-6-31 (April -May 1973). 6. Baugher, D.M. must be approached with con- and Gallace, L.J.; "Methods and Test solder system. The pellet area is coated tages, Procedures for Achieving Various Levels of Power with a protective, high -purity silicon resin siderable care. Overall system reliability Transistor Reliability ", RCA Technical Publication ST- 6209 (Sept. 1973). could deteriorate as a result of exposure before the unit is sealed. An internal view 7. Balan, I.; VanHalteren, C.J.; and Weier, R.M.; "Perfor. of the system is shbwn in Fig. 10. of the components to excessively high mance of Electronics in the Automobile to Date at Chrysler"; Society of Automotive' Engineers, Paper No. temperatures and the forced miniaturiza- 730131 (January, 1973). tion of assembly methods. 8. Bennett, W.P. and Vara, J.S.; "Power Transistors in Automotive Applications "; RCA Reprint RE -18 -6-12 Future (April -May 1973).

The saturation of the ignition market by Future systems certainly will have more 9. Gallace, L.; "Quantitative Measurement of Thermal Cycl- ing Capability of Silison Power Transistors ". RCA solid -state systems is a foregone conclu- "front end" control features to provide Application Note AN-6163.

DIFFUSED EMITTER EMITTER BASE COLLECTOR SUBSTRATE S

J DIFFUSED BASE n+ n- COLLECTOR LAYER f` \

SUBSTRATE

DIFFUSED EMITTER al DIFFUSED BASE

EPITAXIAL BASE LAYER (nl EMITTER BASE SUBSTRATE EPITAXIAL COLLECTOR LAYER (v) n+ p- n- n+

SUBSTRATE

Fig. 8 (above) - Structure os a conventional n -p -n transistor. (b) the pi -nu multiple epitaxial construction.

xp xD view the high- voltage ignition Fig. 9 (right) - Electric field distribution in (a) an n+ pn- Fig. 10 - Internal of n+ structure, and (b) an n+ p- n- n+ structure. (b) switch.

www.americanradiohistory.com In addition, the filter should produce a Surface acoustic -wave minimum of attenuation (<10 dB) in order filter for to achieve good signal-to -noise performance in the receiver. television intermediate The selectivity requirements shown in

Fig. I are not easily achieved on the frequencies production line nor maintained in the presence of drift or aging. In the past, Dr. J. A. van Raalte because of the FCC policy to avoid allocations of adjacent tv channels in the same metropolitan area, problems In recent years, a surface acoustic -wave (SAW) i.f. filter for use in modern television resulting from adjacent- channel in- receivers has been developed at RCA Laboratories; in addition to being small than the terference were minimized. Even where present i.f. filters, such a filter has more stable and reproducible characteristics since it adjacent channels could be received from requires no alignment. This paper reviews the main features of SAW tv -i.f. filters. different cities, the directionality of a good antenna could be used to enhance the desired signal considerably, relative to the unwanted one. However, due to rapid growth of cable television (CTV) systems with simultaneous transmission on all SHARP selective tuning in television The ability of a television receiver to channels, receiver deficiencies, heretofore receivers is a prerequisite for good recep- ignore all channels but the desired one is ignorable, have suddenly become impor- tion; this is particularly true for cable determined primarily by its i.f. filter tant as studies undertaken by the systems which provide a multitude of response. Most of the desired selectivity is National Cable Television Association available channels. provided by the input filter, whose are making increasingly clear. characteristics are shown in Fig. I: Television channel allocations determined by the FCC provide for 6- Acoustic-wave devices MHz bandwidth at specified frequencies, The picture carrier at 45.75 M Hz should be down 3 dB for proper in a Surface acoustic waves in isotropic which in most instances are only 6 MHz transient response media vestigial -sideband system. were first discussed by Lord Rayleigh in apart i.e., channels 2 through 4, - 1885, but their main interest remained in channels 5 and 6, channels 7 through 13, The color carrier at 42.17 M Hz is normally 3 the field of seismology until quite recent- and uhf channels 14 through 83. By dB down to minimize intermodulation effects between chrominance and luminance ly. In 1960 Boemmel and broadcasting a vestieial- sideband signal Dransfeld' signals. demonstrated the use of coherent extending from -1.5 MHz to +4.5 MHz acoustic waves in quartz thereby around the picture carrier frequency, a A deep trap (66 dB) is required where the adjacent- channel sound carrier (47.25 MHz) stimulating the more recent interest in signal bandwidth (including sound and is located; another trap is also desirable acoustic -wave devices for communication chroma information) of approximately where the adjacent- picture carrier (39.75 and signal processing applications. The 4.5 MHz is achieved. This type of fre- M Hz) is located. slow propagation Velocity of acoustic quency allocation, which is conservative I he stop -bane response should be as low as waves (typically 1.5 to 12X105 cm! s of total bandwidth, poses more stringent possible to remove adjacent- channel in- - terference; this is especially critical in a CTV about 105 slower than electromagnetic requirements on the front end of a environment ( .- 40 dB). waves) makes them ideally suited for television receiver if it is to avoid delay lines at vhf or uhf frequencies; thus, adjacent- channel interference and The in- channel sound located at 41.25 MHz should be -20 dB down tri minimize beats microsecond delays are readily obtained various types of cross modulation. between chroma ana sound. in small devices ('H cm).

Dr. John A. van Raalte, Head, Displays and Device Concepts Research, RCA Laboratories, Princeton, NJ, received the BS and MS from the Massachusetts Institute of Technology in 1960. Subsequently he carried out his doctoral research under Professor A. R. von Hippel, Director of the Laboratory for Insulation Research at M.I.T. He was awarded the Engineer's Degree in 1962 and received the PhD in 1964; his doctoral thesis was entitled Conduction Phenomena in Rutile Single Crystals. Dr. van Raalte joined the David Sarnoff Research Center of RCA, Princeton, New Jersey, in July 1964, as a Member of the Technical Staff. He has conducted research in the areas of electro- optical materials, lasers, holography, liquid crystals and their applications to display and recording functions. He was awarded a 1969 RCA Laboratories Achievement Award for the development of a novel television projection system. He assumed his present position in 1970 and is currently directing research in new displays, television rf and i.f. circuitry, acoustic surface -wave devices and tv camera tubes. He is a member of Tau Beta Psi, Eta Kappa Nu, Sigma Xi, the American Physical Society, the AAAS, and a Senior Member of the IEEE.

Reprint RE- 20-1 -5 Final manuscript received March 4, 1974.

www.americanradiohistory.com The interdigital transducer, an efficient means for generating and sensing surface waves in piezoelectric media,Z' shifted the emphasis from bulk acoustic -wave -3dB AT 4217 -3 dB AT 45 75 devices to surface acoustic -wave (SAW) devices. Surface waves, having lower -lo velocities than bulk waves, do not couple COLOR PIX to bulk waves; consequently the surface - 20dB wave energy remains confined to a thin ED -20 AT41 25_, surface layer in the substrate.

z (CO- CHANNEL

Q -30 SOUND) 0 Surface waves have several advantages over bulk waves in many applications: - 40 be fabricated by 1= The transducers can well -established planar technologies and -50 require no bonding of the transducer to the propagating medium; the waves are -60 V ADJ. PI X continuously accessible at the surface and AT 39.75 ADJ. SOUND AT 47.25 can be sensed at different locations - this -7 is useful in tapped delay lines or transver- 4 4 47 49 M H; sal filters. Depending on the choice of the propagation losses can be Fig. 1 - Desired response of i.). input filters. substrate, negligible. An excellent review of surface waves and their potential applications was published by R. M. White.4

SAW filters

A typical SAW transducer is shown in Fig. 2a. This interdigital transducer (IDT) consists of a set of conducting SURFACE WAVES electrode fingers alternately connected to either terminal. When a voltage signal is PIEZOELECTRIC SUBSTRATE applied to the transducer terminals, an (LiNbO3) electric field of spatially alternating polarity appears across the surface of the substrate (Fig. 2b).

If the IDT is fabricated on a piezoelectric SUBSTRATE substrate (as the name implies, a Fig. 2 - SAW filter with interdigital transducers. piezoelectric material converts an electric signal into a mechanical stress, and vice versa), then an electric -field- induced 10 stress is generated by the electrode sin Nit-if-tot/to fingers. This stress will propagate, much Rlf S I N7rtf-fotifo - like ripples in a lake, along the surface of the substrate with a velocity that is its 6 characteristic of the material and crystallographic orientation. Since stress is proportional to the electric field, the 4 contribution from each finger pair is proportional to the applied voltage signal fo and inversely proportional to the finger f-fo= N spacing. O 1 I I v \ s f o- X A SAW filter normally consists of two - 2 IDT's (Fig. 2a); the sender IDT converts the electrical input signal into a surface -4 wave which is subsequently detected by the receiver IDT and converted again into FREQUENCY an electrical output signal. The operation Fig. 3 - Frequency response for uniform IDT. is as follows: An alternating signal of frequency w applied to the sender produces at 'ach electrode finger a cor-

www.americanradiohistory.com proportional to the number of fingers N (especially severe for short fingers) have (Fig. 3). This response is simply the to be taken into account.

T Fourier transform of the coefficients an at phase 41. The coefficients depend or a Thus, in practice, it is more convenient to the applied voltage and piezoelectric con- combine a weighted and a uniform IDT; stants of the substrate (which do not vary in this case, their joint response is indeed from finger -to- finger and therefore do the product of their individual responses. not influence the shape of the frequency The uniform transducer with a (sin response) and the finger spacings and X)/ X response can be made broadband (few finger lengths which can be varied fingers) so as to have little effect on the depending on IDT design. overall filter response, and its response can be included in the computation to Quite generally, in fact, the frequency determine the desired response of the response of an IDT and its finger-overlap weighted transducer. pattern are related by Fourier transformation. This is readily made plausible The desired pass band for the television from a consideration of the nature of a i.f. is relatively rectangular, which would transducer. Each pair of fingers launches imply a finger overlap pattern of the form an acoustic wave with amplitude (sin However, truncation proportional to the amount of overlap X) X. of the (sin finite between the two fingers and with phase X) /X pattern to any numher of ripple in the pass band and given by the location of the finger pair fingers causes Fig. 4 - Experimental SAW i.f, filter. band (Gibb's phenomenon) with the T along the transducer. These waves are stop result that overall selectivity is limited to then summed in the receiver transducer. responding mechanical stress that ^'21 dB. To improve the selectivity, a Such a sum of amplitudes and phases can propagates orthogonally to the finger further weighting function is applied to be recognized as a Fourier transform. lengths in both directions. All these stress the (sin X) /X pattern as shown for a waves are added together with a phase typical i.f. filter design (Fig. 4) By the use shift depending on the location of the Filter design of computer optimization and Fast fingers. Thus if we look at the originating Fourier Transform techniques, designed response of the IDT, say, at some point between sender and receiver, we find a The previous section gives a key to the selectivities of 60 dB, or even greater, are stress that is the sum of contributions design procedure for SAW filters. Given possible. from individual fingers with appropriate the desired frequency response, the phase factors: pattern of finger overlaps (weighting) is derived by taking the inverse Fourier Insertion loss is the N transform. This done efficiently on computer by using the mathematical R(w) _ Eacos (I) techniques of Fast Fourier Transforma- An important requirement for the i.f. n=1 tion. filter is low insertion loss in the pass band: Here the coefficients a are related to the high insertion loss reduces the signal level and therefore deteriorates the signal -to- magnitude of the stress produced by the Since both sender and receiver I DTs each noise performance of the receiver. In n"' finger, and the phase q is determined have their own frequency response, their addition to the obvious increase in cost, by the distance between the nth finger and joint response must be determined. Un- the use of amplification to overcome the sampling point. like cascaded electrical filters, the losses is limited by other practical con- response of two weighted IDTs is, in amplification after the i.f. general, not equal to the product of their siderations: In the simple case, where the I DT consists filter cannot avoid the deterioration in individual responses. This is so because of N fingers of equal lengths, spaced noise performance, while amplification the response of an I DT represents the line equally (X /2 center -to- center) and of before the filter is limited by the desire for integral of the stress wave orthogonal to equal width (X /4), the frequency response low cross modulation with adjacent - its direction of propagation. When two (Eq. 1) reduces to the well -known func- channel signals. Although no precise tion weighted IDTs are cascaded, the energy values can be given for maximum i.f. emitted by the sender is not uniformly filter loss, lumped -element filters can losses less than 10 sin X received by each finger electrode of the easily achieve insertion R(w) _ (2) receiver since they have different lengths. dB; any new filter design should therefore X Thus, the response of two weighted also have the lowest possible loss, transducers can only be represented as preferably less than IO dB. Nrr(r -fo) Nrr(a-W) X- _ (3) the double sum of contributions from W each finger in the sender to each finger in the receiver. In practical cases where the An IDT is essentially a bidirectional The resonant frequency fp is determined sender and receiver may each have 50 to antenna. Its impedance can be by the finger spacing: 100 fingers, this double summation characterized by a real component, the becomes a tedious computation, especial- radiation resistance, and a shunt .ro=cu/2rr= V/X (4) ly if second -order effects such as capacitance determined by the IDT reflections due to fingers, energy extrac- geometry and substrate dielectric con- tion from the wave as it travels under the stant; this capacitance is normally tuned and the bandwidth is inversely transducer, and diffraction effects out by means of an inductor. The radia-

www.americanradiohistory.com tion resistance is determined by the IDT Since the resistivity of the electrode fact, any signal generated at the sender geometry and is roughly inversely fingers must be low enough to avoid that produces a voltage signal at the proportional to the IDT area that is, resistive losses, the designer is faced with receiver will cause the receiver to act as a the product of the number of electrode a delicate tradeoff between excessive sender and re -emit the same signal as if it fingers, their spacings, and their lengths. resistance and excessive mass -loading of were the sender IDT. It can be easily Maximum power is coupled into the the surface; no completely adequate solu- shown that the energy extracted from a sender IDT when the source impedance is tion exists. traveling wave depends on the loading of matched to the radiation resistance of the the receiver: if the output load is matched sender. The problem that results from finger to the impedance of the transducer, then a maximum fraction (50 %) of the acoustic reflections can be greatly alleviated by energy is extracted, one -quarter is The power coupled into the sender using a X/8 finger width and spacing, transmitted, and one -quarter reflected. divides equally into two surface waves instead of X /4. In this design, each finger traveling in opposite directions. Unless is effectively split into two fingers, which In a microsonic filter, the signal energy two receiver transducers are used, one on puts the reflections at twice the frequency that is not received should not be allowed each side of the sender, more than half of of interest, thereby usually placing them to be reflected from the substrate edges, the input energy is lost. Moreover, the outside the frequency range of interest. since it would be detected again with a amount of energy absorbed by the considerable time delay. Consequently, receiver depends on the ratio of load The mass- loading the fingers of that the transmitted energy is usually ab- impedance to transducer impedance; the produces small variations in propagation sorbed by means of a lossy coating (wax) output power is maximized under a velocity also tends to delay the acoustic between the transducers and substrate matched condition and, in this case, half wave more in the central region of the edges. the energy can be extracted. Thus there is transducer than near the edges where an unavoidable 6 -dB loss when only one there are fewer finger overlaps. The net The energy (25% under matched load) sender and one receiver IDT are used. result is a diffraction or defocusing of the reflected between the IDTs, however, Additional losses will occur if the sub- acoustic energy (bending of the wave cannot be absorbed since it travels along strate is lossy (this is not the case for front) so that the wave is no longer the same path as the desired signal. This LiNbO, at television intermediate fre- uniformly incident on the receiver. This reflected signal can produce, at the quencies) or due to impedance mismatch. problem can be largely avoided by using sender, a spurious signal that is delayed As is pointed out later, impedance mis- "dummy fingers" in between the elec- twice as long as the main signal and is match at the receiver is normally used to trically active fingers in order to produce minimize adverse reflections and the in- therefore called the double- transit signal. a uniform delay over the wave front of the Its is sertion loss in practical SAW i.f. filters is effect usually not bothersome and acoustic wave. A more current i.f. filter can be cancelled by means. typically 10 to 15 dB. other design incorporating the X/8 finger However, part of this double- transit pattern and "dummy fingers" is shown in signal is again reflected back towards the Fig. 5. receiver (at %a X VI = 1 / 16 of the sender's Reflections original intensity) and will appear at the receiver, triply delayed and indis- Several types of reflections that occur in Triple transit tinguishable from the singly delayed, microsonic filters can pose problems for primary signal. the designer. One type of reflection is Although we have so far talked about a caused by the unavoidable surface sender and a receiver IDT, one can use The effect of the triple- transit signal is loading and shorting that occurs under either transducer as the sender and the twofold. Since the time delay is fixed by each electrode finger; the resulting other as the receiver. Both transducers, the physical spacing between the acoustic mismatch produces a variation like antennae, can be used to send or transducers and the propagation velocity in the propagation velocity and results in receive acoustic waves by means of the of the substrate, it produces a phase delay reflection of some of the acoustic energy. piezoelectric coupling to the substrate. In which is directly proportional to frequen-

l.o

rc a 0.1 /

o 01 o.o 0 1.0 ,o ioo +ZLOAO Zp

Fig. 5 - Experimental IDT design incorporating Fig. 6 - Ratio of reflected power to output power from IDT as function of narrower finger widths and "dummy fingers ". load impedance.

www.americanradiohistory.com cy. Thus a ripple appears on the pass band of the filter. Furthermore, the delayed signal attenuated by 12dB (1/ 16) being indistinguishable from the main signal, produces a ghost on the tv screen corresponding to the relative delay. For instance, a I /8 -in. spacing between transducers will cause a delay of ^-2 µs, which corresponds to a ghost image separated from the main picture by -34 in. on a 25 -in. tv screen. This condition cannot be tolerated.

Although a lossy substrate attenuates the triple- transit signal more than the main signal, this is not a cure in view of the need for low insertion loss. In fact, LiNbO3 was chosen partly because of its negligible propagation loss at tv intermediate frequencies.

The output load impedance, however, can be varied to minimize reflections. In the extreme case where the output Fig. 7 - Experimental SAW tv -i.f. filter shown next to coil /capacitor filter from MAK -001 module which it replaces. transducer is shorted, no voltage signal will appear across it to generate a re- response then prevents the filter from as much as might be supposed, however, radiated signal; all the incident energy achieving high rejection at frequencies because of the variations in the i.f. passes under the receiver without above the pass band where the system amplifier's input impedance as a function attenuation and there is no triple- transit requires good selectivity and deep traps. of signal level. The integrated- circuit signal. However, no energy is absorbed amplifier is designed to have an in- by a shorted receiver and, consequently, Several techniques have been explored creasing input impedance Z;,, under weak this condition is of no practical interest. with some success to reduce the bulk - signals in order to "unload" the filter and wave response in microsonic i.f. filters. peak up the carrier levels. This increased As the load impedance is reduced from These include roughening the bottom impedance presents a more nearly the matched condition, both the ab- the substrate to a larger, matched load to the SAW filter with a sorbed and reflected energies decrease surface, bonding lossy material, or tilting the various correspondingly large decrease in inser- monotonically. In this range of load crystal faces at an angle to deflect the tion loss. Stop bands in this filter were impedances, the ratio of absorbed to generated hulk waves; none of these attenuated -35 to 40 dB, limited by reflected energy increases (Fig. 6), thus techniques, however, is fully adequate. bulk -wave levels. Temperature stability reducing the triple- transit problem. Better means have been demonstrated, proved comparable to that of the present such as the use of multiple transducers so coil- capacitor i.f. filter, and, of course, no A practical dilemma arises because the that surface waves are added in phase alignment is required in production. in the signal is reduction primary while hulk waves cancel, or the steering of to loss in the filter's pass band. equivalent surface waves along the surface to which leads to degradation of the televi- pnysically separate them from the bulk sion picture, at least under weak signal Acknowledgment waves. These techniques, however, are conditions. Nevertheless, the filters are more costly since they require con- normally mismatched at the output to The author acknowledges gratefully the siderably larger substrates. Bulk wave from the members of reduce the triple- transit signal to a point contributions other suppression to 35 to 40 dB has been Displays and Device Concepts where it produces no visible ghost on the the screen. achieved in simple SAW filters, and this is Research Group who contributed to the probably adequate. success of this project: J. G. N. Hender- son, J. H. McCusker, S. S. Perlman, and Bulk waves T. G. Marshall, Jr. Conclusion Surface acoustic waves remain confined A SAW television i.f. filter was designed in a thin surface layer of the substrate References and incorporated into an RCA XL -100 because their propagation velocity is less television receiver. It replaced the first I. H. Dransfeld. than that of bulk waves and consequently ßoemmel, E.: K.: "Excitation and Attenua- filter (six coils and capacitors) of the tion of Hypersonic Waves in Quartz." Phrr. Rev. Vol. 117 mode conversion does not occur. The Mar. 1950) p. 1245. (MAK -001) i.f. module (Fig. 7). Its inser- higher propagation velocity of various 2. White. R. M.: Voltmer. F. W.: "Direct Piezoelectric tion loss ('«15 dB) is somewhat higher Coupling to Surface Elastic Waves." Appt. Phys. Letters, types of bulk waves causes the IDTs to than that of the production filter due to a Vol. 7 (Dec. f965) p. 314. couple to them at somewhat higher fre- 3. Sittig. E. K.: "Elastic Wave Delay Device" U.S. Patent deliberate impedance mismatch in order 3.360,749 (Dec. 26, 1967). quencies, however, usually just above the to suppress the triple- transit signal. 4. White, R. M.: "Surface Elastic Waves." Prot'. IEEE. Vol. pass band of the filter. This bulk -wave Weak -signal performance does not suffer 58. (Aug. 1970) p. 1238.

www.americanradiohistory.com Very high resolution radiometer A.I. Aronson

High -resolution radiometers are the primary sensors used today in environmental and earth -resource observation satellites to provide day and night cloud -cover pictures, geophysical observations of surface temperatures, land and water features, crop, ecological, and many other resource -study applications.

A.I. Aronson, Senior Engineer, Astro- Electronics Divi- sion. Princeton, N.J., received the BEE from Syracuse ELECTROMECHANICAL SCAN- University in 1951. He joined RCA in 1951 and NERS, despite the sometimes archaic transferred to AED in 1958. Since 1964, Mr. Aronson has been responsible for system and equipment engineering scanning techniques used, have largely on infrared radiometers, horizon sensors, and spec- replaced electron- beam -scanned televi- trometers. Mr. Aronson developed a system concept for sion cameras in environmental satellites, a highly reliable precision radiometer using thermopile detectors and supervised the development of a dual - primarily because they produce channel scanning radiometer using a thermistor geometrically registered multispectral bolometer. Mr. Aronson originated the system desigr and supervised the development of the RCA Very High images extending from the ultraviolet to Resolution Radiometer for ITOS D and E from 1967 to wavelengths longer than 20µm a range 1971. Before the infrared assignments at AED, he worked not possible at present with electronic with spacecraft command systems, including major roles in the development of command systems for television. Other advantages of the elec- SCORE, Tiros, and Stratoscope II. Before his transfer to tromechanical scanners are improved' AED. Mr. Aronson worked on one of the first magnetic radiometric accuracy, linearity, tape recorder systems using solid state electronics stability, video synchronizing generators, and various digita and lack of shading in the image. circuits, with special emphasis on the application of transistors as early as 1952. Mr. Aronson has taught courses at RCA in low- frequency transistor circuits, has published eight technical papers, and holds six patents in transistor circuit applications. 10.0 w THIR HRIR SR

Reprint RE- 20 -1 -18 ó Final manuscript received March 25. 1974. a

I.o I 1 1 1 1 1 1 1 1 1 óo SCMR AV. RR wátà VHRR VHRR (..ATS-FI Fy S-192 MSS F VISOR O 0.11964 1966 1968 1970 1972 1974 1976 1978 LAUNCH DATE

Icruql'ur 117r01 S)m re, Ír

IIRIR High resolution infrared Nimbus i;idiumeler

I IIIR temperature humidity infrared Nimbus radiometer SR Scanning radiometer ITOS VHRR Yen high resolution 11OS radiometer S('MR Surface composition mapping Nimbus radiometer VItRR Very high resolution ATS -F

1 I S-F radiometer VISSR Visible-infrared scan radiometer.SMS S-192 S -192 multispectral Skylab scanner \VIIRR Advanced very high I IROS-N resolution radiometer \1SS M oh(upectral scanner FRIS

Fig. 1 - Evolution of resolution capability of imaging radiometers on NASA spacecraft (for infrared channel only).

www.americanradiohistory.com Table I - Typical performance parameters of the VHRR.

Parameter Typical performance

Orbit altitude, nmi 790 ± 25

Resolution

IFOV, mrad 0.6

Modulation transfer function @ 0.6 mrad (square wave) 0.4 IR, 0.5 visible

Spectral coverage. gm Fig. 2 - Very high resoluiton radiometer, two views. Visible 0.60 - 0.70

Infrared 10.5 - 12.5 VISIBLE PHEAMPLIFIF:R Sensitivity

Visible SNR (p- p /rms) @ 65 It 4.0 END HOUSING IR minimum resolvable temperature (MRT) @ 3(5 K 0.3 SIGNA L ELECTRONICS Dynamic range MAIN HOUSING Infrared. K 180 - 315

Visible, fL 65 - 10,000

Electrical bandwidth, kHz 35

Scan rate, scans /min 400

In -flight calibration

SCAN DRIVE 4 and 300 ELECTRONICS Infrared, K

Visible Sky, solar target

Fig. 3 - Exploded view of VHRR. Scan linearity ±0.05 mrad jitter along any vertical straight line

Power. VV 6.5 The Very High Resolution Radiometer oriented, three -axis- stabilized sensor Size, in. 18.4 x 10.77 X 12.75 (VHRR) starts a second generation of platform of the spacecraft. The high performance infrared scanning overriding emphasis throughout the Weight, lb 21 radiometers, as shown in Fig. 1. The development of this sensor has been radiometers of the 1964 -1972 time frame design simplicity, low power consump- had subpoint ground resolution in the tion, and compact size. range of four to five miles; the VHRR launched on October 15, 1972 provides This two -channel instrument can provide 0.47 -mile resolution. This performance day or night cloud -cover pictures in the breakthrough, in the infrared band, was 10.5- to 12.5 -µm infrared band and sunlit - This halves the spacecraft transmission made possible by the use of a mercury - pictures in the 0.6- to 7 -µm visible -light bandwidth requirement. In the backup cadmium- telluride detector cooled to ap- band. The instantaneous field of view mode, either one of the two radiometers proximately 105 K by a passive radiator (IFOV) of each channel is 0.6 mrad, can be used to provide two parallel cooler. The VHRR made the first in non -multiplexed providing a subpoint resolution of 0.47 channels of data operational use of this new passive fashion with some degradation of the nmi. Cross -track scanning is produced by radiator cryogenic technology - an im- received S. N on the ground. portant technology which represents the a single 45- degree mirror, rotated at 400 r/ min by a direct -drive (gearless), only practical means at present for The high -resolution performance of the brushless, dc torque motor; spacecraft reliable infrared detector cooling for VHRR infrared channel is obtained using orbital travel provides the along -track periods of a year or more in space. a cooled mercury -cadmium- telluride picture component. Table I lists the key detector. Cooling the infrared detector to parameters of the VHRR. 105 K provides the sensitivity (D *) and required to achieve very VHRR design fast response The two V H RR's are operated in a right - high resolution. Cooling is accomplished The VHRR shown photographically in and left -hand configuration in either a with a unique, passive cooler which uses Fig. 2 and in an exploded view in Fig. 3, is primary or backup mode. In the primary geometry, special surface treatment, and a scanning radiometer designed for use mode, both instruments are used with staging to better radiate heat to space, on the ITOS D series spacecraft. Two their mirrors phased 180 degrees apart to while rejecting thermal inputs from the VHRR's are mounted on the earth- time multiplex infrared and visible data. earth, the sun, and the spacecraft.

www.americanradiohistory.com 8E ir- 0111-TE- COEE, RADIATOR [\ L -N07 Only two interfaces in I COOLOLER the VHRR optical I /// system are critical with respect to dimen- 11111 DC DETECTOR S I / ++ftllllll' RESTORER I is RELAY OPTICS 1 sional stability. One the spacing BUFFER L J ¡ AMPLFIERS between the relay lenses and the infrared detector; the other is the spacing between

" VISIBLE" DC O the primary and secondary in the ICBROIC DETECTOR RESTORER LOW BEAM telescope. The relay /detector assembly is SPUTTER NOME AMPLIFIER maintained at the detector temperature of CLOCK 16V I05 K and is inherently stable. The effects SYNC DC -DC 6V COMMANDS TELEMETRY GENERATOR F- CONVERTER of temperature variation on the telescope -24.15V -l6V L are minimized by using an Invar tripod assembly to mount the secondary with

Fig. 4 - Simplified block diagram of VHRR. respect to the primary. The variation in the spacing for a -5 to +45° C temperature change, using Invar, is ±0.000167 in., resulting in a shift of the The primary optics of the VHRR is a lenses, spectral filters, and detectors. The primary focus of only ±0.0075 in. This Dall- Kirkham 5 -in.- diameter telescope. optical design stresses simplicity and a focus shift is negligible, since the depth of Secondary optics are used to spectrall} minimum of optical elements. This is focus at the image plane is ±0.125 in. separate the infrared and visible channels particularly true in the visible- light- and to relay them to the respective channel optics, where the detector is the A typical VHRR subsystem has asquare -. detectors. field stop and the only image- forming wave MTF of 0.5 in the visible and 0.4 in element is the telescope objective. The the infrared tegions, with a variation of In -orbit radiometric calibration targets infrared channel design minimizes the no more than 5% in MTF over a are used to monitor the sensitivity of each effects of cold-plate/ detector movement temperature range of -10 to +40° C. spectral channel for possible changes by collimating the energy between the with orbital conditions and time. The radiometer housing and the cooler. Fig. 5 orbital calibration references in the in- shows the layout of the optical system. Detectors frared band are space and the radiometer backscan black body, both of which are The information bandwidth and sensitivi- The 5 -in. Dall- Kirkham viewed once per revolution. A two -point objective com- ty requirements of the VHRR dictate the bines an ellipsoidal primary calibration reference in the 0.6 to 0.7 µm and a use of a fast, high D* (>1010cm Hz"' spherical secondary, and is the band is obtained by scanning space and a simplest watt-1) detector. The only commercially form of a mirror sun -illuminated, diffusely transmitting two- telescope that meets available infrared detectors that satisfy target. the modulation transfer function (MTF) these requirements and can operate at requirements. An 0.018 -in.-diameter temperatures (100 ±10 K) achievable with silicon photodiode located A block diagram of the VHRR is shown at the focal a passive radiative cooler, are mercury - in Fig. 4. point of a 29.6 -in.- focal -length objective cadmium- telluride (HgCd -Te) and lead - in results an instantaneous field -of-view tin- telluride ( PhSnTe). At the present, (0) of 0.6 mrad. An infrared channel f- HgCd -Te is superior to PbSnTe with Optical design number of f/ 0.8 chosen to maximize the respect to responsivity and D* and was 0.89 system S/ N requires a detector size chosen for the VHRR application. The The VHRR optical subsystem consists of of 0.0026 in. to provide a 0 equal to 0.6 characteristics of the VHRR infrared a scanning mirror, a Dall- Kirkham mrad. The infrared relay magnification is detector are summarized in Table II. telescope, a dichroic beamsplitter, relay then 0.018/0.0026 or 6.92. The visible channel uses a selected version of the Hewlett- Packard stock 5082 - 4204 19.69 PIN silicon photodiode.

MCT DETECTOR (FIELD STOP) BRUSH LEBE 10.6 - 12.6 NIs MOTOR FILTER SCAN DANE Electronics

The VHRR signal electronics shown in the block diagram of Fig. 4 is mounted on four PC boards in the electronics SILICON PHOTODIOI assembly box. The scan -drive electronics (FIELD STOP) are mounted both inside the scan drive and on a board mounted in the scanning and optics housing, as shown in Fig. 3. DALL K.IRKHAM The visible-light preamplifier is a PRIMARY OPTICS separate assembly, mounted near the Fig. 5 - VHRR optics, physical locations of components. silicon photodiode.

www.americanradiohistory.com

¡MPUFIER Signal electronics channels, are added to structure the Table II - Infrared detector characteristics. complete video format.

The signal electronics consists of the c harairri.,rtr analog electronics required to amplify The sync generator also provides the 5- and condition the signals from the in- kHz scan -drive reference and the phase - Responsive element size. in. 0.0026 X ±0.0003 frared and visible -light detectors. modified 5 -kHz required when time - Dctectivity, D (Am°,. 3000, 60°) multiplexing two radiometers. In this cmHz°`W' 1.6 X 10°1 (at 105 K) The infrared detector forms one leg of a mode, one radiometer slips phase until its bridge circuit. The bridge resistors are index pulse is 180 degrees out of phase Bias current (chosen for max DA). mA 55 selected by calculation to provide the bias with the other radiometer index pulse. required for maximum detector D* and Bias power. mW 52 mW to balance the bridge. The bridge is Rcsponsivity. R. ; > 7500 at A,.,,, and at optimum bias followed by an amplifier whose noise current (at 105 K) figure is approximately 1.5 dB. This Scan -drive electronics Spectral characteristics I. 1 lµm (0 at operating temperature requires a dc- restore function to provide a each of which is switched four times per zero radiance reference and dc stability. revolution, is controlled by Hall elements Stability farter 55 °C soak) .1 resistance < 10% < mounted in the stator. The signals from signal 10% or.R. . noise ..20% or ..1D*,, 20ah1 The dc- restore function is provided by three Hall elements are amplified and fed feeding back offset errors to balance the to a matrix, which decodes six states from bridge when the radiometer views zero the three inputs. The six commutation radiance (space) once per mirror rotation. on the six at the signals turn motor drivers channel enabled. The radiometer with IR appropriate angular positions of the enabled is the slave and shifts phase with The visible -light channel analog elec- rotor. respect to the other (master) radiometer. tronics consists of a current -mode The phasing logic compares the time preamplifier. With this arrangement, the An incremental tachometer. consisting of delay between a master once -around effective diode load resistor can he made a 750 -line encoder, provides velocity feed - pulse and a slave half-around pulse. and very high (5 megohms in the VHRR) to hack for servo control of scan velocity generates an acquisition signal if the two provide both high S/ N and good linearity and jitter. The encoder rotates between a motors are not phased within a 300 ± 100 of signal output vs. brightness. The dc- gallium arsenide diode and a ,us acquisition window. restore function, once per revolution phototransistor, providing a sine -wave when the scan mirror views space, is used output whose frequency is proportional These two signals are used to control the to maintain low- frequency stability and a to velocity. The velocity signal is shaped 5 -kHz velocity reference, which will cause radiometric reference. and squared and fed to a phase detector. the slave motor to either slow down or The phase detector saturates to provide speed up, depending on the phasing of the maximum drive until the motor comes two motors. A signal from the timing within the lock -in range of the phase Sync generator logic occurs at a 194 -Hz rate, with a detector. The phase detector output then period of 3.41 ms. and is used to control oscillates between saturation and cut -off The sync generator electronics provides the number of 150 -kH7 clock pulses fed to until frequency coincidence is obtained all the timing signals required by the between the velocity track signal and the the 5 -kHz velocity counter. This is ac- radiometer. complished once every 3.4 ms by remov- 5 -kHz clock reference frequency. At the point of frequency coincidence, the ing one clock pulse of the 150 -kHz clock if Timing is initiated by the index pulse the slave is too fast or by adding one clock velocity loop is locked in phase. After generated once per mirror rotation by the pulse, if the slave is too slow. This results lock -in. motor jitter results in a pulse - encoder in the scan drive. This analog in a modified 5 -kHz pulse that is one width- modulated. 5 -kHz signal at the pulse is reclocked to a 5 -kHz scan phase -detector output, the instantaneous period of the 150 -kHz clock earlier or reference pulse to eliminate any scan jitter in time. This causes slip modulation being proportional to the later the motor to from sync and all other timing signals in scan -drive jitter. The pulse-width- phase ±6.6µs periodically at a 3.41 -ms the video. Timing of all events is done by rate. Thus. the scan phasing changes 1.94 modulated signal is filtered, frequency - counting down from the 300 -kHz compensated, and amplified to drive the ms/ s for a maximum phasing time of spacecraft clock and decoding event motor. 38.7s for 180 degrees. times. The decoded events activate a multiplexer to gate a seven -level stairstep reference voltage. 0- and 6 -volt levels for The VHRR phasing logic will cause the sync pulses, temperature telemetry, and scan mirrors of the master and slave infrared and visible video derived from radiometers to rotate 180 degrees out of Scan drive on -board calibration targets. phase. This phasing operation occurs automatically when both radiometers are The drive mechanism is shown in Fig. 6. Precursor frequency bursts, which are operating, one with the IR channel The scan mirror is attached to a stainless - unique to the infrared and visible -light enabled and the other with the visual steel shaft. which is supported on a pre-

www.americanradiohistory.com loaded (5 -Ib) pair of 102H angular con- CULU The cooler cross section is "L" shaped, HONEYCOMB PLATE tact ball bearings. Labyrinth seals are PANEL with sides, to form a "scoop" type of mounted on each side of the bearings to SECOND i Cc geometry surrounding the cold plate. SURFACE MIRRORS i retain lubricant. An optical encoder disc KAPTON Both the first -stage cooler and the outer BAND having a 750 -line -pair servo track, SUSPENSION frame have this same scoop configura- INFRARED SYSTEM together with a once-around pulse track, FILTER tion. This geometry is unique among AT 240°K is mounted to the motor shaft. The SPACE contemporary spacecraft passive cooler encoder sensor and emitter module is DETECTOR AND RELAY designs, all of which use a conical con- OPTICS ASSEMBLY mounted to the housing, accessible for figuration to shield the cold plate. The I HONEYCOMB alignment. A one -piece stainless -steel PANEL earth -facing, or scoop part of the cooler COLD PLATE . housing locates the stationary com- shields the cold -plate from thermal inputs ponents and attaches by a flange to the r from earth albedo and self- emitted in- scan and optics housing. frared energy. No additional shielding FIAS 4.` OUTER HOUSING -- from the sun is required, because the LOW CONDUCTIVE STANDOFFS cooler is mounted on the side of the Cooler spacecraft opposite the sun and the cold EARTH plate is shielded from the sun by the The shown cooler, in the photograph of spacecraft and the sides of the scoop. The Fig. 2 and in the cross -section view in Fig. Fig. 7 - Cross section of VHRR passive radiator cooler. cold plate radiates heat to space. It 7, is used to cool the infrared detector to absorbs heat from the low- emittance, approximately 105 K. The cooler vapor- deposited -gold scoop sides, ther- employs a combination of geometry and mal conduction from wires and the cold - special thermal surfaces to shield it from The cooler consists of three main parts: I ) plate support, and radiative coupling (and reject thermal input from) the a cold plate, 2) a first -stage coolers, and 3) from the back of the cold plate. The spacecraft, the sun. and the earth, while an outer frame. The cold plate is thermal- thermal equilibrium of these inputs and simultaneously heat radiating to cold ly isolated from the inner frame by a the radiation to space result in a "natural" space to accomplish this completely unique truss -type Kapton band suspen- cold -plate temperature of approximately passive cooling function. sion. 95 K. The cold plate is electrically heated to 105 K with a proportional-control heater circuit in order to maintain a constant detector temperature if the natural cooler temperature should vary within the orbit or with time.

Radiative inputs to the cold plate from NY LA6D T LUBRICANT the first stage are minimized by main- RESERVOIR taining the first stage at approximately et 160 K. This cooling is accomplished by use of second- surface mirrors mounted MIRROR on the vertical part of the first stage. INCREMENTAL LABYRINTH SEAL TACHOMETER Second- surface mirrors have an ab- I ` PRELOAD SPACED DUPLEX PAIR BEARING sorbtivity ( a) of 0.07 and an emissivity (e) MOTOR of 0.8, which make them both good reflectors of earth albedo and good emitters to cold space. Fig. 6 - Scan drive. sectional view. The first stage is surrounded by the outer frame, which is thermally isolated from the radiometer housing and the DAYS FROM LAUNCH 40 80 120 160 200 240 280 320 60 400 440 480 spacecraft by a Delrin support bracket. 10 The outer frame is cooled to ap-

6 proximately 240 K by means of second - 7 surface mirrors mounted on its earth - facing side. 4 q w

2 All internal cooler parts are gold- coated, and the spaces between the outer frame, 0 600 1000 1600 2000 2500 3000 3500 4000 4500 5000 5500 6000 ORBITAL REVOLUTION inner frame, and cold plate are insulated with multilayer gold -on- Kapton thermal Fig. 8 - Thermal performance of cooler in orbit. blankets to minimize radiative coupling losses.

Performance of the coolers on the

www.americanradiohistory.com The VHRR tracks cold and warm eddies to yield more accurate local surface thermal patterns than is obtainable with aircraft, as an aid to the Bureau of Fisheries. Using VHRR's, NOAA is currently mapping three US river basins operationally for snow-cover extent.

In meteorology, the higher resolution and multispectral capabilities of the radiometer allows improved analysis of both local and synoptic atmospheric data.

And finally, an interesting proposal for a feasibility study concerns the use of VHRR data to monitor ground temperatures and find areas conducive to the hatching of screw worms. The area would then be "seeded" with sterile male screw -worm flies to reduce the screw - worm population. Fig. 9- NOAA- 2 satellite VHRR pictures of Hurricane Ava taken June 1973. (Picture on left imaged by VHRR infrared channel: picture on right imaged by visible channel.) Acknowledgments

NOAA -2 meteorological satellite since 1972. As of the date of this paper, the two Many people at RCA, the Goddard launch is shown in Fig. 8. VHRR's, operating continuously since Space Flight Center of NASA, and the launch, have logged a total of 25,500 National Oceanic Atmospheric hours of successful operation. Administration contributed to the Contamination protection success of the VHRR. At RCA, recogni- The cooler operation has been particular- tion for original engineering concepts -ice trap is incorporated in the An anti ly noteworthy, as shown in Fig. 8. This during the early VHRR development cooler to prevent the contamination of shows the cold -plate heater power re- should be given to J. Armington, R. (105 K) optics from icing or cold quired to maintain the detectors at 105 K. Williams, D. Binge, P. Roehrenbeck, E. outgassing- product accretion. The margin above the natural cooler Goldberg, and T. Furia. Significant con- temperature is found by dividing the tributions to the management, design, works by surrounding the cold The trap cold -plate power in milliwatts by 2.4. fabrication and testing of the VHRR inner tube, which is at detector Degradation in 16 months of operation during the prototype and flight stages with a 240 K tube to temperature, outer has been small and has leveled off, were made by G. Barna, F. Eastmen, R. prevent direct contamination of the cold indicating a VHRR -cooler lifetime that Lambeck, F. Lang, C. Masucci, A. relay optics. This arrangement is shown will exceed the instrument design lifetime Mayo, R. Scott, and L. Weinreb. in Fig. 7. A foreign molecule will freeze or of one year by a large margin. condense at the base of the cold inner The prototype and flight phase of the tube before it can work its way between The high -resolution, low- noise, and VHRR program was funded under a the inner and outer tube and contaminate jitter -free pictures derived from these portion of NASA Contract NAS5- 10306, the cold relay optics. In addition, heaters instruments can be seen in the pictures of with J. O'Brien as technical officer. are incorporated on the inner frame (first Hurricane Ava shown in Fig. 9. stage) and the cold plate to heat these parts to about 240 K, on command, to purge these elements if accretion of con- taminants reduces the margin between Bibliography the cold -plate control temperature and VHRR data utilization the "natural" temperature of the cold I. Hudson. R.D., Jr., Infrared System Engineering, John plate. The improved resolution of VHRR data Wiley and Sons. New York (1969). has already aided the scientific studies of 2. Schnapf. A., The Evolution of ITOS, the Operational Environmental Satellite of the 1970's," Ninth Space Con- ice, sea surface thermal patterns and gress. Cocoa Beach, Fla., April 19 -21, 1972. -Observations current, snow, and the atmosphere. 3. from the Nimbus I Meteorological Satellite;' American Geophysical Union, Seattle. Washington. Dec. VHRR performance 29, 1964, NASA SP -89. The advantages of high -resolution view- 4. Aronson. A.I.. "Application of Infrared Sensing in NEREM 70 Record. The NOAA -2 (originally ITOS D) ing of ice at night has found application in Meteorological Satellites;' - 5. topper. M., Johnson. D.S., "Toward an Operational meteorological satellite, carrying two ice mapping during polar nights, support Weather Satellite System." Astronautics and Aeronautics. VHRR's, was launched into a circular, for the US Navy weather operations, and Vol. 3. No. 16. (June 19651. 6. Breckenridge. R.W.. (labran. F.. "Cryogenic Cooling near- polar, 790 -nautical mile orbit from ice mapping in Alaska and the Great Systems for Spacehorne Experiments" Proceedings of the Western Test Range on October 15, Lakes. E lectro- Optical Design Conference, 1972.

www.americanradiohistory.com High- capacity high- data -rate instrumentation tape recorder system O. E. Bessette

A 240 -megabit /second serial- bit -stream recording system using a longitudinal (fixed - head) magnetic recording technique called HDMR (High Density Multitrack Recording) Fig. 1 - 164 -track high- density multichannel recording head. has been developed. This system provides maximum bits per square inch of tape at reliable in -track packing densities. Unique "unitized" fabrication techniques have been used to density multitrack recording (HDMR) construct single -stack magnetic heads (record /play on the same head) at track densities technique has been developed. HDMR of over 100 tracks per inch. Commercially available tape is accommodated by the use of provides maximum storage of data bits error detection and correction. HDMR technology, applied to the implementation of a on a minimum quantity of tape (in other typical ground instrumentation recording system, allows key performance parameters of words, maximum bits/ square inch). To 240 Mb /s serial data input, 108 in. /s tape speed, a 142 -track head, 240 -Mb /s serial data achieve this goal, both the in -track pack- output, and less than one bit error in one million. ing density (or bits/ inch along tape), ano the track density across tape (tracks/ inch) have been increased. HDMR is unique, however, in the degree of extension of track densities. This SENSOR OUTPUTS whether syn- adjustments; 5) is the most easily en- feature permits extended area packing density at reliable recorded wavelengths. chronous scan (television, radar, electro- crypted form of data; and 6) can be An alternate approach, which has been optic) or non -formatted (i.t., rf, tele- computer compatible. widely promoted, uses standard inter - metry) are basically analog at the data - range instrumentation group (IRIG) source. However, because of the relative Until recently, A/ D converters were track format heads with in -track densities ease of handling data in digital form, the limited to low bandwidth and /or low extended up to 35,000 b/ in. This ap- trend is to use A/ D converters dynamic range applications. With the proach yields an area packing density of to digitize data from these sensor sources advent of A/ D systems capable of ade- 0.980 X 106 b /in.' with an IRIG track for processing. Typically, a digital format quately processing such complex signals density of 28 tracks /inch. Experience has l) provides minimal as broadcast -quality color tv and high performance shown that sensitivity to head -to -tape degradation during data processing, mul- resolution radar, a need for recording and separation losses at such short processing tigeneration copying, or editing; 2) allows digitized data at rates in the wavelengths results in high bit -error rate easy detection and correction of errors hundreds of megabits per second has (BER), low reliability of the head -to -tape and dropouts; 3) accommodates very emerged. high data rates through parallel processing; 4) drastically reduces operational To address this need, a longitudinal high Oliver E. Bessette, Ldr. Aerospace Systems GCS Recording Systems, Government Communications and Automated Systems Division, Camden, N.J. received the BSEE in 1958 from the Worcester Polytechnic Institute. He joined the RCA Defense Communications System Division and participated in the Design and Development Training Program. In 1959, he joined the Speech Engineering section of CSD which is now the Recording System group. As a junior and then senior engineer, he participated in recording systems hardware design and development in areas such as wideband fm modulation /demodulation, phase-linear fm processing, low -noise wideband amplifiers, magnetic head design, tape- transport development, servos, rotating bandwidth, automatic time base correction (ATC) using electronically variable delay lines (EVDL) and digital encoding. In addition to circuit and subsystem design, his assignments included program management, cost control, customer liason and field support activities for many types of RCA ground based, airborne and space video and wideband instrumentation recorders. In 1969, Mr. Bessette was promoted to Engineering Leader and has since been the Principal Investigator in development of a new product line of RCA recording systems called High Density Multitrack Recording (HDMR). This IR &D activity has led to development techniques which have proven the realizeability of digital recording systems that can store digital data in magnetic tape memories at 3 = 10" bits /reel of tape, with transfer rates up to 300 megabits /second. Mr. Bessette is a member of IEEE group on Magnetics and holds a patent on a magnetic head.

Reprint RE- 20 -1 -21 26 Final manuscript received December 18, 1973.

www.americanradiohistory.com HIGH RATE HIGH RATE system are shown in Fig. 2. Although this ANALOG DIGITAL DATA DIGITAL DATA ANALOG INPUTS f OUTPUTS diagram shows an analog -to- digital converter at the system input, the digital A/D SWITCHER SWITCHER D/A signal to be recorded can actually be derived from various sources:

DEMUX MUX

} A serial or parallel digital input EDAC EDAC ENCODERS CORRECTOR A digitized analog input A "self -test" signal

DATA DESKEW ENCODERS DEFLUTTER } The selected input data is demultiplexed

RECORD DATA to many parallel lines and parity checked EQUALIZERS DECODERS AMPLIFIERS by the EDAC encoder. All data and EDAC NRZ (non- return -to -zero) signals MAGNETIC PLAYBACK HEADS AMPS are phase encoded and equalized for optimum record bandwidth. Each is then amplified to a suitable Fig. 2 - HDMR system block diagram. channel record level and applied to the multitrack magnetic head. interface, and compromised tape usabili- latter effect is the reason wide tracks are ty. Hence, in development efforts to normally used. An error detection and obtain high packing densities (b/ in?), correction (EDAC) system for HDMR The reproduce portion of the HDMR maximum effort was devoted toward has been developed which effectively the same HDMR head to extending the track density while main- system utilizes circumvents the tape dropout errors, read out the stored data. Each head taining reliable in -track density. thereby providing a BER of better than output is amplified, filtered, and limited 10-6 at packing densities of over 2 X 106 Successful video head technology, by analog circuitry. These circuits and the requires very developed by the Broadcast Engineering b /in.2. [The EDAC system final stage of the record amplifier are the Group and RCA Laboratories, led direct- little overhead (implementation), does only analog circuits associated with the ly to unique fabrication techniques for not require that the data be blocked or basic HDMR record /reproduce system. ultra -high- track -density longitudinal formatted, and can be utilized to correct These multichannel Circuits contain only heads. Applications of this technique errors generated during data processing twenty active devices per channel and have successfully demonstrated track external to the recorder.12] With this require only one adjustment: record densities of more than 80 tracks per inch, system, standard instrumentation tape current. No playback equalizer is re- yielding an area packing density of over can be used direct from the factory quired since this function has been 2X106 b /in.' with an in -track density of without burnishing and cleaning. The replaced by a simple and unique equalizer only 25,000 b/ in. HDMR technique thus provides a tape on the record side. Thus, the normal utilization capability that allows the complications associated with equalizer Typically, the two -inch, 164 -track recording of ultra -high data rates at adjustment have been eliminated and the reduced HDMR head (shown in Fig. 1), operating reasonable tape speeds by spreading the analog reproduce electronics are to 50% of the normal requirement. at 30 inches /second, records at the same high rate signal over many parallel tracks data rate as a 28 -track IRIG version on tape. The development of high density operating at 120 inches /second, but the heads was the prime element in HDMR limiter are decoded HDMR version requires only two -thirds technology. Other important elements The outputs of each independently of clock circuitry. During the in -track density. This means a more contribute substantially to the perfor- decoding, clock information is extracted reliable head/ tape interface plus in- mance and reliability: creased tape utilization, i.e., less tape from some of the data channels and used used, or more record time available. to strobe the data out of the decoders and Heads - No gap scatter, 1000 hours life; into the deskew /deflutter buffers.' The lack of scatter" in HDMR heads (no For example, a requirement to record 60 Head -tape interface - Commercial tape, "gap reliable in -track packing densities, accurate interchannel- time -displacement error Mb /s for 1 hour can be satisfied by tracking; between adjacent tracks) allows this HDMR with a 4500 -ft roll of tape. IRIG BER Error detection and correction; minimal implementation of clock cir- format at 35,- to 40,000 b/ in. requires - h /in. Double density code, unique cuitry. Typically, only one clock is needed 18,000 ft of tape (a reel greater than 16 in. equalization, high track density, incremen- to strobe out ten tracks, the limit being diameter, with a dreaded tape splice). tal operation; determined by the amount of dynamic Simple electronics - No equalizer to adjust, skew. A VCXO (voltage controlled only 20 active elements per channel; and Experience has also shown that 28 crystal oscillator) or an external reference Deskew No overhead, no static skew, simple tracks provides "overkill" in signal - - can be used to control the data output /inch system to -noise ratio even at high b/ in. figures. A stability. The source selected serves as the 5 -mil (0.005 -inch) track operating at 25,- reference for the recorder capstan and the 000 b/ in. reliably produces a measured output clock for the deskew /deflutter HDMR BER of less than 1 X 10-8 if tape system buffer. Digital time base correction is imperfections are ignored, and 1 X 10 -4 accomplished in the deskew /deflutter with tape imperfections included. The The major elements of a typical HDMR buffer with a single digital integrated

www.americanradiohistory.com Table I - Typical specifications - HDMR -240G. 60

Data rate' Record 40 to 240 Mb /s continuously variable SO Reproduce 40 to 240 Mb /s fixed, selectable Data format Serial NRZ -L in/out

Auxiliary data inputs Up to 12 lines @ 2 Mb! s digital (or other signal formats)

Record time 23 min. @ 240 Mb's 138 min. @ 40 Mb! s

Start time 50 ms. record 100 ms. reproduce

Error rate 2 x 10 ° maximum 20

Magnetic tape 2 -in. on NAB reel 16 -in. (12.500 -ft) maximum qs i0 q' IOs 10 lape interchangeabilityTape recorded on one HDMR -240G can be BIT ERROR RATE reproduced on a second HDMR -240G

Packing density In -track density of 20,000 b/ in. 1.4X 1066 /in.' Fig. 3 - Signal -to -noise ratio (peak /rms) as a function of bit -error rate for 7 -, 8 -, and 9 -bit quantization (for rms S/N subtract 5 dB).

1 -otal storage Primary data 3.3 X 10" bits Capability Auxiliary data 3.3 X le hits Head life 1,000 hrs. (random access high to process serially. In Size 65 ft.' circuit memory) which this mode, the replaces all the functions normally ac- HDMR system would play back a suf- Power 115 Vac. 60 Hz, I phase, 2 kW

Number of tracks 112 data, 14 EDAC, 12 aux complished by sophisticated analog time ficient number of bits to load a memory X Ficad stacks One record, play" base correctors such as MATC, CATC, (e.g., 1.2 106 bits), then stop, rewind a and EVDL. short distance to index Head construction Alfecon 11 Unitized HDMR the following data, and wait for another playback Track width 0.008 -in. spacing 0.014 -in. The coherent parallel data from the command. Bit error rates of 10 -4 to 10-7 Iransport Vertical with waist -level tape loading, buffers are error checked and corrected are typical of standard HDMR systems. vacuum bins for fase start %stop by the EDAC system, and multiplexed This range encompasses the requirments lape speed 107 in./ s at 240 Mbis (18 in. at 40 Mb's) for D/ A conversion or further digital for most system applications. For in- processing. Cabinet Welded EMI design on casters, transport door is stance, NTSC broadcast video, which has pressurized to keep tape compartment clean been A/ D converted with 8 -bit Special effects Edit, auto search and other special quantization4 and recorded, will produce features also available Applications a reconverted (D /A) video signal with a 47 dB signal -to -noise ratio (S /N) for 'Also available at 120 Mb: s on I -inch tape The HDMR system is available in one - BERs less than 10-5 (refer to Fig. 3). The "Read while write available head stacks, plus increased and two -inch versions. Both versions use (two basic HDMR system operates at 240 X de -skew electronics) standard instrumentation tapes and have 106 bits /second. This is considered a a 2:1 data rate ratio. The one -inch version typical bit rate for today's needs. has exactly half the signal processing electronics and records up to 120 Mb/ s, or exactly half that of the 240 Mb /s In addition the system reflects the version. HDMR applies equally well to capabilities of present A/ D converters ground, airborne and space data storage and covers the required rates for present requirements. Low power, small size and and near future systems. HDMR weight and high reliability are ac- provides the capability to record a serial complished by simplified channel elec- 240 -Mb /s bit stream as might be tronics and transport design. An available from a very wideband data link, aerospace version of HDMR can be or to record multiple, synchronous data implemented in a configuration weighing outputs (typical of A/ D converters) at less than 250 pounds and occupying less line rates of 240 / N X 106 bits /second for than 5 cubic feet, storing and producing each of N lines. An illustration of such a digital data at a rate of 250 Mb /s, and 240 -Mb /s recorder/ reproducer is shown requiring less than 350 watts of power. in Fig. 4 and typical specifications are listed in Table I. HDMR provides computer compatibility for analysis of high speed and wideband References

data. Eight or sixteen output lines may be I. Melas, C. M.; "A New Group of Codes for Correction of chosen, with output rates of 4 X 106 dependent Errors in Data Transmission ", IBM Journal (Jan. 1960) pp. words/ second or less, for dumping into 58-65. 2. Griffin, J. S (RCA); "Ultra High Data Rate Digital

present minicomputers or main -frame Recording ", Proceedings of SPIE Conference 1973 , input buffers. Incremental reproduce Dayton, Ohio. 3. Thompson. C. R. (RCA) and Hayes. J. M. (NASA - capability is available with HDMR to tioddard). "High- Data -Rate Spacecraft Recorders ", allow data reduction/ processing at Proceedings NTC Conference. 1972. Houston, Texas. 4. Viterbi. A. J.; Principles of Coherent Communications Fig. 4 - HDMR system configuration. transfer rates which are normally too (McGraw Hill, New York; 1966) p. 257.

www.americanradiohistory.com bandwidth capability, especially at the Linear transistor upper useful frequency end of a particular transistor. The drawbacks are lower overall efficiency and the requirement power amplifiers that the transistor must tolerate forward Dr. E. F. Belohoubekl D. S. Jacobson biasing at high cw dissipation levels without thermal runaway.

New microwave transistors using emitter ballasting provide high gain, linear operation Over the last few years, transistors have with power outputs of 2.5 W at 2 GHz and 700 mW at 4 GHz. With suitable circuit been developed that include heavy techniques, these devices permit the development of octave bandwidth power amplifiers emitter ballasting and thus permit class -A that are in many respects superior to present TWTs. operation up to 4 GHz. Some of the possible applications for these devices in the communications field are very wide - band driver amplifiers with large dynamic range for laser communications THERE IS a general trend today to now make it possible to develop wide - links, truly linear power amplifiers for replace existing TWT amplifiers by solid - band class -A microwave power multiple carrier repeater stations, or state devices. Most solid -state amplifiers amplifiers. "'' Class -A operation has very wideband i.f. driver amplifiers for for present communications applications definite advantages compared to stan- millimeter wave upconverters. Many are relatively narrowband and operate dard class-C operation; besides a much other applications probably will open up under saturated conditions, with class -C larger dynamic range, it offers lower when system designers become aware of transistor amplifiers dominating the second -harmonic output, lower the possibilities of class-A power lower microwave frequency range. Re- amplitude modulation/ phase modula- amplifiers in the microwave frequency cent improvements in bipolar transistors tion conversion, higher gain, and a wider range.

In this paper we discuss the present state

Erwin F. Bslohoubek. Head. Microwave Circuits David S. Jacobson, Mgr.. RF and Microwave Devices of the art in ballasted microwave tran- Technology, RCA Laboratories. Princeton, N.J., re- Design, Solid State Division. Somerville, New Jersey. sistors and their applications to -Ingenieur received the BChE from the Polytechnic Institute of ceived the degree of Diplom in 1953 and the Some PhD in Electrical Engineering in 1955 from the Technical Brooklyn in 1954. He then joined the General Electric microwave power amplifiers. University in Vienna, Austria. From 1953 to 1955, he Company where he participated in the Chemical and results obtained recently with FET tran- worked as Research Assistant at the Institute for High - Metallurgical Training Program. After a tour in the U.S. sistors are included to illustrate the up- Frequency Techniques of the same university. He joined Navy, Mr. Jacobson joined the Semiconductor Division the RCA Tube Division in Harrison. N.J., in 1956 and of Texas Instruments in 1958 and worked on processing coming full band coverage with linear transferred to the Microwave Applied Research techniques for grown -junction silicon transistors. Mr. amplifiers to X -band. Laboratory in Princeton. N.J., in 1957 asa Member of the Jacobson joined the RCA Semiconductor and Materials Technical Staff. In this position. he worked on the Division in 1961 as a device development engineer: he development of magnetrons. electrostatically and engaged in the development of germanium double - and germanium and gallium - magnetically focused traveling -wave tubes, and a cross- diffused transistors Linear microwave power tran- He Project ed field microwave delay tube. In 1969. Dr. Belohoubek arsenide tunnel diodes. was promoted to Design sistors was put in charge of a group working on microwave Leader of the Varactor Device and Deve opment hybrid integrated circuits. In this capacity. he is respon- group where his major responsibilities included the varactors for the sible for the development of passive and active MIC development of high -reliability LEM The present state of the art of class -A circuits which include high -power transistor amplifiers. program. Mr. Jacobson transferred to the RF Transistor Design group in 1965 where he worked on the 2N3866 power transistors at frequencies above 1 multipliers, linear bipolar and FET amplifiers, electron - he was responsible for beam, semiconductor amplifiers, circulators and overlay transistor. Subsequently. GHz is illustrated in Fig. I. The TA limiters, and small microwave subsystems. He received the design and development of microwave powe, transistors. including the 5 -watt, 2 -GHz 2N5921 and 2N6266 numbers represent RCA developmental an Outstanding Performance Award of the RCA Elec- and the 2 -watt. 2 -GHz 2N5920 and 2N6265. Mr. acob- tronic Components Division in 1963 and an RCA devices operating common base in son was promoted to Engineering Leader of the Laboratories Achievement Award in 1967. Dr. Transistor Design group in 1968. In that Belohoubek holds five patents and has written more than Microwave was responsible for the design and develop- twenty papers in the areas of microwave techniques. position he transistors. low -noise tran- traveling -wave tubes, and microwave integrated circuits. ment of microwave power and transistors for CATV applications. He is He is a Senior Member of the IEEE. sistors. presently responsible for the design of all RF and Microwave 'transistors and Microwave Integrated Cir- RCA del cuits. TA6170 (104

RCA (124el TA575e 2e) (70e) EXPERIMENTAL RCA 80-1i6 r4a81 GaAs FET 2115920 (2031 (edB)

4 node) (606) NP 35830*

EXPERIMENTAL GaAs FET (60111 dB)

2 3 4 5 6 7 e 9 IO FREQUENCY - GPIs

Fig. 1 - Power output curves (above 1 GHz) of several linear power transistors, including RCA developmental devices.

Reprint RE- 20 -1 -22 Final manuscript received November 12. 193 29

www.americanradiohistory.com microstrip carriers. The results for the EMITTER BOND MD HP device are also for common base operation, while the MSC device performance is given for common emitter. The gain (shown in brackets) of the latter WE OXIDE r... device would probably increase %/ somewhat if used in a common base vi i G L'''''''''''''' .- microstrip carrier. Common base is used i i nearly exclusively with all class -C tran- EMITTER WE WIDTH sistors in the microwave range; it also EASE EPITAXIAL LAYER provides higher gain performance in \\\\\\\\\\\\\\\\\\\\\\ \\\\\ class -A but requires careful circuit design SUBSTRATE \\ to avoid instabilities and makes the dc bias circuitry more complex. Fig. 3 - Cross section of ballasted overlay transistor.

The effect of ballasting on class -A tran- drive levels, the transistor can be biased at ballasting may become excessive. For sistor performance can best be un- much higher collector currents without high- frequency operation, a technique derstood by comparing Figs. 2a and 2b. incurring thermal runaway. In the ex- designated as emitter site ballasting is The transistors in both cases use the same treme of very high drive into saturation, utilized where the contacting geometry of geometry, an overlay structure with the average dc power dissipation the metallization is designed so as to place broad, thick emitter metallization. The capability approaches that of class -C the polycrystalline layer as ballast resistor unballasted version (Fig. 2a) shows a very operation, which for this particular tran- in series with each emitter site. The rapid fall -off in gain for collector currents sistor is approximately 12 watts. TA8758 that provides linear performance above 200 mA. The onset of this decline is at the I -watt level at 3.4 GHz is a typical this type characterized by a thermal imbalance in The key to successful true class -A opera- example of of ballasting. the chip, where one part of the transistor tion at high-power levels lies in heavy hogging starts the current while the rest of enough ballasting of the emitter to prac- Linear power amplifiers the transistor practically is turned off. If tically eliminate the termal imbalance operated at currents considerably above within the transistor chip. Fig. 3 shows Besides providing linear multiple -signal the point where the rapid gain decline sets the crossection of a typical overlay tran- amplification, ballasted transistors in in, the transistor will be destroyed by sistor, which includes emitter class -A operation generally also offer local overheating. The gain fall -off is not ballasting. A polycrystalline silicon layer improved high- frequency response and influenced only by ballasting but also by interspersed between the aluminum wider bandwidth capability than their magnitude rf At high the of the drive. metallization and the shallow diffused class -C counterparts. emitter sites acts as the vehicle for in-

20 troducing ballasting to the emitter sites. Class -C operation is severely limited by For heavy ballasting, the metallization of turn -on problems at high frequencies due each emitter finger is interrupted where it high Q SED RF AND DC BIAS to the of the input impedance, normally connects to the common - which makes it very difficult to match w 2N5921, - . V 15V emitter bond pad. The current has to pass into the transistor over a wide -frequency Io f 2 GHz through the polycrystalline silicon range. Class -A operation provides more layer, which permits heavy emitter gain, and since the transistor is always debiasing in the order of 200 mV. The biased well into forward conduction, it debiasing provides degenerative feedback does not depend critically on a good input to prevent excessive current in any por- match to turn on. On the contrary, for 200 400 COLLECTOR CURRENT -mA tion of the transistor and thereby controls truly wideband operation, the input im- the hot spotting. The effect of the emitter pedance may be even negative over a Fig. 2 - Effect of ballasting on class -A transistor finger ballasting is illustrated in Fig. 2b. performance:2a (curve above) for unballasted transistor; and 2b (curve below) for ballasted transistor. No rapid gain fall -off exists up to 700 mA of collector current, except for a gradual decline associated with the normally observed fr (gain- bandwidth product) 20 TA el TO fall -off at high collector currents.4 At 600 VA15v ImW mA, a power output of 3 W with 33% P.-- efficiency can be obtained.

-- I-- If the transistor size is rather small, hot -'-- vk33x -_ spotting generally is less pronounced and 300mW PO .3W lower levels of emitter ballasting are i sufficient (e.g., 40 to 100 mV). This is particularly important if operation at 0 higher frequencies is desired where the 100 200 300 400 500 600 700 Fig. 4 - Hybrid integrated- circuit, 1 -2 GHz power COLLECTOR CURRENT -mA-- reduction rf gain of due to heavy amplifier.

www.americanradiohistory.com large part of the operating bandwidth. To 20

overcome cascading problems under IB these conditions, two class -A amplifiers

are generally combined by 3 -dB hybrids 0.1 in each stage, which not only provides 14 more power but also effectively decouples i TA 8758 the stages from each other. 2 Ve 15V INPUT . ßi7 OUTPUT Ie200mA I 4 GHz z 10 DISTRIBUTED LINE The above techniques were successfully ó BOND demonstrated in the octave bandwidth L- 6 WIRE VOS band power amplifier shown in Fig. 4. 0.001 Voc 5V Amplifiers with 1 GHz or more V. -05V bandwidth capability are required, for tef 130mA example, in gigabit data links. To obtain Pm,c 60 m IMD >20dB AT P.c. 40mW approximately IOW of cw power, two 0.1 I 10 100 PHASE LINEARITY .1 3.5 directly parallelled power transistors are POWER INPUT -m 45 5 5.5 6 in employed each arm of the balanced FREQUENCY -GHz output stage. Power sharing is ensured by Fig. 6 - Performance of emitter- ballasted transistor individual dc biasing of each of the four amplifier at 4 GHz. Fig. 7 - Performance of two -stage FET amplifier. output transistors. The near -saturation performance of the three -stage amplifier, For frequencies above 4 GHz, GaAs microwave power transistors makes it which measures 5.65 x 1.5 x 0.925 inches, field -effect transistors are presently being possible to develop wideband amplifiers is shown in Fig. 5. The overall efficiency developed that have superior high - that in many cases represent a superior of the amplifier including bias network frequency capabilities and better power - alternate to present TWTs. Except for a losses ranges from 16% to 24% over the sharing characteristics. The positive lower tolerance to high- temperature band. A single power supply of 17 volts temperature coefficient of GaAs makes operation, the solid -state amplifiers have drives the amplifier module. the FET much less susceptible to thermal the inherent advantages of better runaway conditions prevalent in un- reliability, smaller size and weight than The same technique that was used ballasted bipolar transistors; parallelling even the most advanced minitubes, much successfully to combine four transistors of many cells is thus possible without hot simpler power requirements, better gain in the power stage of the above amplifier, spotting. The FET has, however, a more flatness, less susceptibility to mechanical if applied to the smaller high -frequency difficult topology since all three elec- shock and vibration, and they require no transistor TA8758, would provide a 2 -W trodes of each cell appear on top of the warm -up time. Within the next 1 to 2 power output capability for a linear chip. Interconnection of many cells years, bipolar class -A amplifiers with 3- amplifier in the 3.7 to 4.2 GHz com- without incurring excessive parasitics W output are expected to reach 4 GHz, munications band. Fig. 6 shows the and feedback is thus more difficult than while FETs should approach the I -W performance of a single -cell transistor at 4 with bipolar transistors. Recent ex- level at IO GHz. GHz. The P/ Pat curve shows a gain perimental results obtained with power reduction at low rf drive levels, which FETs are included in Fig. 1. The highest Acknowledgment indicates a small remaining thermal non - power output to date is 0.8 watt at 4 GHz, uniformity throughout the chip. The which was achieved with six parallelled The authors gratefully acknowledge the amount of this gain compression depends cells on a single GaAs chip.' contributions of many colleagues, in par- on the degree of ballasting, the location of ticular, R. Duclos and T. Boles in the area the bias point and, since the local hot Preliminary results from a two -stage FET of ballasted bipolar transistors, and of A. spots generally are also affected by the amplifier covering the frequency range Presser in linear -amplifier design. The output loading, to some extent on th. from 4.5 to 6 GHz are shown in Fig. 7. power FETs were developed by L. Napoli load impedance. This amplifier uses two chip- mounted, and A. Dreeben, and the FET amplifier experimental, single -cell FETs in a circuit results were supplied by G. Theriault and that measures only 1/2 X 3/4 inch. Simple, R. Tipping. combined, distributed, and lumped excellent gain 10 matching networks provide ln° flatness and phase linearity. FETs have References 8- the added advantage of a more linear

Pow characteristics close to the satura- I. Presser, A., et al. "1.5 Watt. S -band Linear Transistor á 6 tion level compared to that of bipolar Amplifier ". High - Frequency Generation and Amplification Pic 20mW Conference, Cornell. pp. 445 -450, June 1971. VB17V transistors. Therefore, the intermodula- [c OUT 2. Presser. A., et al. "1 -2 GHz High -Power Linear Transistor 4 IT28A tion distortion of FETs is generally lower Amplifier, RCA Review. vol. 33, pp. 737 -751. December Te 30C 2N5'20 TA8170 at power levels a few dB below saturation. 1972. 3. Renkowitz, D.. "L- and S -Band MIC Power Amplifiers Come of Age ". NEREM Record. vol. 13. 1971. 4. Muller. 0., "411 tralinear UHF Power Transistors for CATV Proc. IEEE. vol. 58, pp. 1112-1121. July 010 12 14 16 1.8 Applications'. 1970. FREQUENCY - GHz. 20 Conclusion \.Ipoli, L. S.. et al, "High -Power GaAs FET .Amplifier- A \l eltigate Structure ", Digest, IEEE International Solid Fig. 5 - Performance of 1 -2 GHz power amplifier. The advent of well -ballasted bipolar .Slate Circuits Conference. pp. 82 -83, February 1973.

www.americanradiohistory.com Digital image transform M PELS - yt encoding W. B. Schaming

N FELS The transmission and storage of continuous -tone imagery in digital form is a subject of intense investigation by a large and ever increasing number of researchers. One particular area of investigation receiving much attention is image encoding. In general, the purpose of this research is to determine ways in which an image can be encoded to reduce the total PELS number of bits required to represent the image. Most techniques produce a small amount of degradation to achieve large reductions in the number of bits required. Obviously, the n FELS more degradation one can tolerate, the greater the bit compression ratio (BCR) that can be Fig. 1 - Partitioning of NxM array of pets into nxn achieved. Two measures of this degradation are the mean -square error in the subpictures. reconstructed image ana subjective ranking by untrained observers, although no one asserts that either of these measures is adequate as an image quality standard. Many different approaches to image encoding have been studied, including one- and two - dimensional differential pulse -code modulation (DPCM), one- and two -dimensional n=16 are adequate and yield comparable transform coding, contour coding and a host of variations. The subject of this article is the results. It is felt that 8X8 is large enough research carried on in Advanced Technology Laboratories in the area of two -dimensional to provide flexibility in the coefficient digital image transform coding. thinning studies and yet small enough that real -tirng implementation is not ruled out. However, if the additional storage is acceptable, 16X16 subpictures tend to produce better results at a given W. B. Schaming, Advanced Technology Laboratories, bit rate. Camden, NJ, received the BSEE in 1962 from Gannon TRANSFORM coding is generally un- College, where he was the recipient of the alumni award derstood to mean that the picture for general scholastic excellence. In 1964, he was in the spatial awarded the MSEE from the University of Pennsylvania. elements domain are Subpicture transformation Mr. Schaming joined the Astro- Electronics Division in mapped via some linear process into the 1962, where he worked on slow -scan television cameras transform domain. When the Fourier The Fourier transform has been the for space- observation systems. In 1964, he transferred to transform is used, the transform domain the Advanced Technology Laboratories, where he has vehicle for coding studies under the been engaged in projects requiring computer solution is the familiar frequency plane. The current project. However, the transform and simulation. The projects have included signature elements in the transform domain may is applied to image data in such a way that analysis, part retrieval systems, Monte Carlo techniques, then be thinned by some algorithm and generation and processing of magnetic anomaly signals, the input data to be transformed exhibits generation of buffered I/O systems, generation of quantized. This thinning and quantiza- fixed, known symmetries. graphical output, functional simulation of computer tion leads to the reduction of the number systems, and real-time processing. Mr. Schaming has of bits required to represent an image, many years of experience in computer systems and The Fourier transform and cor- programming, with primary emphasis on the integration since some of the transform coefficients responding inverse given by Eqs. 1 and 2 of digital computers into engineering systems. In addi- are discarded and most of the retained can be used to transform each image tion, he has conducted extensive research and develop- coefficients may be represented by fewer This is ment in the areas of image encoding and picture subpicture. discrete transform processing. bits than an original PCM sample in the equivalent to a Fourier series expansion, spatial domain. Various transformations, and will therefore treat each subpicture including Fourier, have been compared being transformed as though it were a in detail by other researchers. (For an period of a two -dimensional periodic excellent discussion of transform picture function as shown in Fig. 2. coding the reader is referred to an article by Wintz,' which also provides an extended bibliography on the subject.)

R Image partitioning

An image represented as an NxMarray of R (pels) be picture elements can partitioned SUB - 1. PICTURE into nxn subpictures as shown in Fig. R R BEING R R TRANS- NXM image contains NMI n` (nxn) sub- FORMED pictures. R

The choice for the size of subpictures has R IS A REPETITION OF THE SUBPICTURE been varied in our studies. As indicated in R BEING TRANSFORMED Ref. I, subpicture sizes between n=4 and

Reprint RE- 20-1 -10 Fig. 2 - Image subpicture shown as 2 -d periodic Final manuscript received December 17, 1973. function.

www.americanradiohistory.com J-. functions and at the same time guarantee- symmetric about both the x and y axes. bll b12 b13 b14 ing that phase information is preserved When the original 4X4 subpicture of Fig.

It exactly. 3 is transformed, 16 independent b21 b22 b23 b24 numbers are required to specify the sub-

b31 b32 b33 b34 picture. When the 8X8 even symmetric Given the 4X4 image subpicture shown in subpicture is transformed, still only 16 b43 b41 b42 b44 Fig. 3 whose elements are the brightness independent numbers are contained in values b;;, an 8X8-element subpicture can the 64 Fourier coefficients. This is Fig. 3 - 4x4 subpicture taken from image. be constructed as shown in Fig. 4. (A 4X4 because the transform of a function initial subpicture is used here for ease of which is even symmetric in both illustration.) With the x and y axes drawn directions produces coefficients which N-1 N-1 as shown in Fig. 4 (no samples on either are pure real and equal in all quadrants. F(u,v) _ - 2_, 2_, f(x,y) axis) and the brightness samples assumed Since the transform of the even N X=o v=0 to be in the center of the sampling symmetric subpicture is pure real, the exp {(-2rri/N) (xu+vv)} (1) interval, the 8X8 subpicture is even phase is zero and need not be considered. N-1 N-1 f(x,y) = - 2_, 2_, F(u,v) N u=0 v=0 exp {(2n-i/ N) (ux+vy)} (2) tin b11 b12 b13 b14 b14 b13 b12

022 b21 b23 b24 b24 b23 b22 b21 Edge effects b31 b32 b33 b34 b34 b33 b32 b31

When viewed as a periodic function, b41 b42 b43 b44 b44 b43 b42 b41 differences in brightness values on opposite edges of the subpicture create b41 b42 b43 b44 544 b43 b42 b41

1331 discontinuities in the periodic function. b31 b32 b33 034 b34 b33 b32 The existence of these discontinuities contributes high frequency energy in the b21 b22 b23 b24 b24 b23 b22 1321

transform domain. Anderson and Huang b11 b12 b13 014 1314 b13 b12 b11 suggested increasing the subpicture size by one row and column whose values are the average of opposite edge subpicture Fig. 4 - Even symmetric 8x8 subpicture constructed from 4x4 subpicture of Fig. 3. samples as one way to help smooth this discontinuity.2 They observed improvement by using this technique.

b21 1321 1321 b22 b22 b22 Phase considerations b23 b23 b23 b24 b24 b24 The transformation of an 8X8 subpicture produces 64 complex coefficients in the Fourier domain. However, since the im- Fig. 5 - Subpicture samples from Fig. 3. Fig. 7 - Subpicture samples as appear in the even symmetric subpicture of Fig. 4. age samples are real, the complex coefficients exhibit conjugate symmetry through the origin. Therefore, only half of the samples are independent, and an PERIOD M PERIOD 8X8 subpicture can be completely b21 specified by 32 coefficients (64 numbers). b21 1 b21 b22 1322 b22

Research by Andrews and Tescher of 1323 b23 b 23 USC indicates that the phase information b24 b24 b24 is extremely important.' In fact, when encoding the amplitude and phase of the complex coefficients, more bits were Fig. 6 - Subpicture samples in the periodic representation of Fig. 5. assigned to phase than to amplitude.

Symmetric transformation of PERIOD PERIOD PERIOD subpictures I b21 b2 1 b21 b21 b21 b21 1,22 b. b22 1322 b22 b22 bz3 1,23 023 b23 b23 1323 A method which has been the subject of b24 b24 1324 1324 b24 b 24 our investigations is presented here for preventing the warped spectrum caused by discontinuities in the periodic Fig. 8 - Subpicture samples in the periodic representation of Fig. 7.

www.americanradiohistory.com 8 x 8 EXPANDED TO 16 x 16 SYMMETRIC TRANSFORM 100 parable variances.

Thus it seems reasonable that all a 99 8 x 8 TRANSFORM coefficients within band of equal manhattan distance have equal significance, on the average, in the image reconstruction. 98

Thinning techniques 97 Some of the most popular techniques for the thinning of coefficients are the follow- 96 ing. The simplest approach is to keep the first n coefficients (ordered by variance) in the transform plane. No additional 95 bookkeeping is required since the same

1 1 coefficients are t 1 I I I I I I l I I 1 I transmitted for every 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 subpicture in the image. ri is determeined NO. OF INDEPENDENT COEFFICIENTS by examining the variances of the coefficients. The next approach in order Fig. 9 - Convergence of two transform methods. of complexity is to keep the 3] largest coefficients (ordered by their variance) in Furthermore, the discontinuity in the required to represent an image comes each subpicture. This requires additional periodic function is eliminated, thereby about through the thinning and quantiz- bookkeeping bits to indicate which removing extraneous high frequency ing of the coefficients in the transform coefficients are being transmitted. Both energy in the spectrum. Suppose that the domain prior to transmission. Many of these approaches can be made to adapt original samples b21, b22, h23, h24 have differenct approaches have been in- to local picture content by letting o vary brightness values as shown in Fig. 5. The vestigated and reported in the literature. for each subpicture by thresholding the same line of samples in the periodic The approach used in our investigations amount of energy preserved. This re- function would appear as in Fig. 6 which is a combination of these techniques. quires that 77 be also transmitted for each clearly shows a discontinutiy. On the Some of the variations introduced were subpicture. other hand, the same samples when taken dictated by our initial goal of having the from the expanded 8X8 even -symmetric encoding process useful for real time techniques subpicture of Fig. 4 appear as in Fig. 7. implementation. This implies that the Quantization The periodic rendition of this same line in entire image is not available prior to the the symmetric subpicture appears ás encoding process. The simplest quantizing scheme is one in shown in Fig. 8 which clearly has no which each coefficient is normalized by discontinuity. Therefore no articficial its standard deviation (a) such that all high frequency energy is inserted into the Ordering of coefficients coefficients can utilize the same quantizer specturm. and are assigned the same number of bits. Most of the reported coefficient thinning Another approach is to normalize the techniques require that the coefficients be coefficients by their a, but vary the Comparison of symmetric and ordered by the variance computed from number of bits assigned to the nonsymmetric subpicture all the subpictures in the image. Our coefficients. The number of bits assigned transformation finding in the limited experiments per- can be made proportional to the formed has been that this type of coefficient variance. The series representation of the image coefficient ordering agrees quite well with taken from the even -symmetric the ordering of coefficients by their dis- transform will converge faster than that Strategy in current experiments tance from the dc coefficient. As shown in taken from the straight transformation. Fig. 10, all coefficients in a band are an To illustrate this, a test picture was Two approaches have been used in this equal manhattan distance from the dc transformed first by directly transform- project, one of which is adaptive, the coefficient C,,.,,. [The manhattan distance ing each 8X8 subpicture in the image to other is not. Both approaches use the is defined as the number of increments in obtain the coefficients for each subpic- same quantizing technique but different the .v and y direction from C5,o to The ture. It was also done by transforming thinning techniques. The test image was variance of any coefficient taken a 16X16 symmetric subpicture derived partitioned into strips of subpictures as from the subpictures in the image tends to in Fig. 1 I. a 256 X 256 image, form each 8X8 region of the image. The shown For with increasing band number, there convergence rates for the two methods decrease are 32 strips each having 32 8X8 since the energy content of most images subpictures. Each strip of subpictures can be seen in Fig. 9. decreases with increasing spatial frequen- was encoded independent of all other cy. This seems to be supported by strips. This approach io. dictated by our Transform coefficient thinning the findings of Landau and Slepian.4 assumption that the entire image is not and quantization Furthermore our observation has been stored prior to encoding. Each 8X8 ele- that all coefficients within a band of equal ment subpicture in a strip was then The reduction of the number of bits manhattan distance tend to have corn- expanded to an even -symmetric I6X16

www.americanradiohistory.com subpicture and transformed. The assuming eight bits for each a and six bits Overhead = (8X8+6)/(32X64) + (3/ 64) variances of the coefficients shown in Fig. for the dc mean. = 0.081 bits/ pel (3) 10 from all subpictures in the strip where then computed. However a separate variance for each coefficient was not In the adaptive thinning and quantizing Generation of 6 -bit quantizer computed. Instead, one variance was scheme, a value of a for each of the eight computed for each band of coefficients bands must be transmitted for each strip The 6 -bit quantizer used in our ex- having the same manhattan distance of subpictures, in addition to the dc mean periments assumes that each coefficient from Co,o. In our investigations we never since that is the maximum number of has a gaussian probability density and use more than the eighth band of bands which any subpicture might re- has been standardized prior to quantiz- coefficients when using 8X8 subpictures, quire. In addition, a 3 -bit number is ing. The cut points were generated first therefore only eight a's are computed. transmitted for each subpicture in the and the representative values for the For example a3 corresponding to band 3 strip to specify the number of coefficient reconstruction levels were taken as the is computed form the values of Co.2, CLI. bands being transmitted. Therefore the midpoint between successive cut points. C2.o taken from each of the 32 subpictures required overhead for the adaptive The cut points were generated using the in the strip. This is justified since our technique assuming eight bits for each a method described by Andrews which observations show that coefficients in and six bits for M0,0 is given by minimizes mean -square error.' bands of equal manhattan distance tend to have comparable mean and variance. After computing the a for each coefficient COEFFICIENT BANG band, the coefficients are then divided by the proper a and passed through the same 6 -bit quantizer. (in the case of the dc coefficient, C0.0, the mean is subtracted before normalization.) All six bits are not utilized for each coefficient band. A typical assignment of bits for the coefficients in the first eight bands is given in Fig. 12. When a 3 -bit quantizer is desired for band 3, the three most significant bits are retained from the 6 -bit quantizer.

Nonadaptive thinning has been employed Fig. 10 - 64 independent real coefficients from transform of even -symmetric 16x16 subpicture. by retaining the first N bands of coefficients in each subpicture. Adaptive thinning has been used in which the 1ST STRIP OF SUBPICTURES number of bands varies for each subpic- 2ND STRIP OF SUBPICTURES ture. Successive bands are kept until a preset percentage of the total energy in the subpicture is exceeded. Only the determined number of bands of coefficients are then quantized and transmitted. For this adaptive approach a minimum of three bands and a max- LAST STRIP OF SUBPICTURES imum of eight bands are used when the subpicture size is 8X8 elements. Fig. 11 - Strips of subpictures from original image.

Overhead bits COEFFICIENT BAND a For the nonadaptive thinning and _quantizing scheme using N bands [N(N+1)/2 coefficients] for each subpicture, the required bookkeeping bits include those required to code the dc mean, Mon, as well as the value of a for each band. This data is required for each strip of subpictures. Therefore the overhead per picture element is given by

NX8 +6 Overhead - bits/ pel (3) 32X64 Fig. 12 - Typical bit assignment for first 36 coefficients.

www.americanradiohistory.com Effects of bit errors Simulations have been performed in- of code words, a bit error in the most serting bit errors into the transmitted significant bit of a code word near the One of the concerns in any encoding data stream using both 8X8 and I6X16 peak of the distribution results in a process is the sensitivity of the encoded subpicture sizes encoded with the same maximum shift of 32 reconstruction data to channel noise which results in bit average bit rate. levels. For example, an error in the most errors in the received data stream. significant bit of code word 32 results in Transmission links are often the code word 0, which is located at the characterized by specifying the probabili- Although the total number of bits re- left end of the distribution. Likewise, an ty of any bit being in error at the receiver. quired to describe each of these images is error in the most significant bit of code This probability may vary 10-2 from to the same, and the number of bit errors word 3 I results in code word 63, which is 10 5. inserted in each image is the same, the at the right -hand end of the distribution. resulting images appear very different. It appears that simply revising the order The effect of bit errors on the Using an 8X8 element subpicture size, of the code words on the left half of the reconstructed image received over a there are 1024 subpictures in the total distribution would provide considerable transmission link greatly depends upon image. There are only 256 subpictures in reduction in error sensitivity. This results the encoding process. When transmitting the total image when the subpicture size is in the inverted order code words also an image in the spatial domain using a 16X16 elements. As a result, when using shown in Fig. 13. For this ordering of the PCM sample for each picture element, the smaller subpicture size, the chances of code words, a bit error in the most the bit errors appear in the reconstructed getting a bit error in the most significant significant bit of a code word near the image as the familiar salt- and -pepper bits of very low frequency coefficients is peak of the coefficient distribution results noise. When DPCM (delta modulation) four times as great. in a small shift in the reconstructed level. is used, the bit errors will appear as A bit error in the most significant bit of a streaks in the image. When a transform code word on the tail of the distribution encoding process is utilized, however, the Images were encoded using a natural will now result in a shift to the opposite results are quite different. A bit error will ordering of the code words in a six -bit tail. However, the probability of being on result in the wrong value for a coefficient quantizer as shown in Fig. 13. When less the tail of the distribution is much in the transform domain. However, the than six bits are used for a give'. co- smaller. inverse transformation operation will efficient, the desired number of high - spread the effects of the bit error over the order bits are retained. In studying Fig. entire subpicture to which the coefficient 13, one can conclude that the natural belongs. The resultant degradation will ordering of the code words is disadvan- depend upon the subpicture size used in tageous. The code words near the peak of Experimental results the encoding process as well as the the coefficient distribution have the allocation of bit patterns to the highest probability of being used. The images shown in Fig. 14 are the result reconstruction levels in the quantizer. However when using the natural ordering of applying the techniques described

STANDARDIZED COEFFICIENT DISTRIBUTION (ASSUMED GAUSSIAN)

RECONSTRUCTION A A A A A A A AA LEVELS

NATURAL ORDER O 2 30 51 32 33 62 63 6 BIT CODE WORDS

INVERTED ORDER

31 30 29 I O 32 33 61 62 63 6 BIT CODE WORDS

Fig. 13 - Two possible code -word assignments for 6 -bit quantizer.

www.americanradiohistory.com above to the sample image. The pertinent Table I - Reconstructed image statistics. data applying to each picture is given in Table 1. The percent mean -square -error

values in the table are computed from the Picture Subpicture No. of bands Bit error Bit compression % mean reconstructed image by number size or % energy rate Bit /pe! ratio square error

N N I 0 1 Original NA 0 8.0 J + image N2 L/ (Ii; - r9)2 2 8X8 5 bands 0 0.8 10.0 0.56 r=1 j=1 3 16X16 10 bands 0 0.8 10.0 0.50 % mean square error= N N 100% 4 16X16 8 bands 0 0.58 13.79 0.70

1 5 8X8 99.3% 0 0.77 10.4 0.54 L, (/ )2 (5) 6 8X8 99.5% 0 0.85 9.4 0.48 N2 pi 7 8X8 5 bands I0 -' 0.8 10.0

8 16X16 10 bands 10' 0.8 :0.0 0.81 9 16X16 10 bands 10' 0.8 10.0 1.77 where r,; are reconstructed image 10 16x16 10 bands 10-' 0.8 10.0 1.54 samples.

Notes: - For 8X8 subpictures the number of bits /coeff. in successive bands in 6,5,4,3,2, Conclusions - For the apaptive images (5 and 6) the number of bits / coeff. in successive bands in 6,5,4,3,3,3,3,3, - For 16X16 subpictures the number of bits / coeff. in successive bands is 6,6,5,5,4,4,3,3,3,3, Picture 10 is identical to 9 except that the inverted ordering of the code words described in Fig. 13 is The results produced by these ex- used. periments indicate that a reduction of Mean square error extremely large due to large dc errors in many blocks 10:1 in the number of bits required to represent an image, with only small degradation, is feasible using these techniques. The effect of a noisy of a tv -type signal. 2. Anderson, G.B. and Huang, T.S.; "Piecewise Fourier transmission channel, although different to -frame correlation Transformation for Picture Bandwidth Compression," from PCM or DPCM systems, is not This would be equivalent to a three - IEEE Trans. Comm. Tech., Vol. COM -19, No. 2 (April 133 -140. catastrophic and, in some cases, is less dimensional transform. 1971) pp. 3. Tescher, A.G.; "Adaptive Intra -and Inter-Frame Image disturbing to the observer. Coding Utilizing Fourier Techniques," USC Image Process- ing Research Semiannual Technical Report No. USCEEE 444. pp. 24 -33. Although no tests have been attempted, it References 4. Landau, H.J. and Slepian; "Some Computer Experiments he take an in Picture Processing for Bandwidth Reduction," Bel! may possible to additional System Tech. Jour.. (May -June 1971) pp. 1525 -1540. in the frame direc- transform -to -frame I. Wintz, P.A.; "Transform Picture Coding, "Prot IEEE, Vol. 5. Andrews, H.C.; "Computer Techniques in Image Process- N.Y.; Academic Press; 1970) p. 144. tion to take advantage of the high frame- 60 (July 1972) pp. 809 -820. ing" (New York,

Fig. 14 - Images corresponding to data in Table I.

www.americanradiohistory.com Numerical NUM ERICA I. CONTROL (NC) is a control technique for the control of production tools by information prerecorded in sym- applications for spacecraft bolic form. It is a method designed to optimize in advance and then control the components work of a metal- cutting machine tool.

D. Charrier I M. F. Luft The NC concept

RCA Limited has been involved with numerically controlled (NC) equipment since 1969. The concept of NC was developed during Originally, its function was to provide an in -house means of machining high -tolerance the late 1940's by John Parson, an parts for Canadian communication contracts. The business has since expanded to include engineer who was preparing templates for the fabrication of satellite components required for commercial applications. Resulting aircraft component construction. The from this growth, enhancements to the in -house NC shop have doubled the value of the complex airfoil design was defined equipment (present value approximately $1 million) and increased capacity by 400/o. This geometrically by a series of curves; this paper is divided into three sections: 1) definition of NC, 2) computer interface, and 3) approach enabled Parson to represent present and future usage at RCA. integral template points with a table of X and Y coordinates. His men drilled holes in the template sheet metal which formed straight lines, tangent to all the various points on the desired curve -a complex calculating and measuring task in itself. An error in any one of thousands of measurements might ruin a template. The operators then filed the remaining cusps until a faired curve resulted. Parson conceived the idea that the dimensions could be recorded on punched cards; that M. Luft, Management Information Systems, RCA Dan Charrier, Fabrication Model Shop, RCA Limited. the cards could be read by a Limited, Montreal, graduated machine; from McGill University in Montreal, graduated from Ecole Polytechnique in 1970, and, that 1968 with a B.Sc. degree in mathematics. He then obtaining his Bsc A and Mechanical Engineering degree. servo- motors could move the entered the University of Waterloo to obtain a Masters He then joined Canadian General Electric's technical template to the correct position under the degree in Mathematics and Computer Science. Mr. Luft support staff specializing in the area of NC applications. cutting tool. spent one year as a programmer analyst with the Federal In 1972, Dan developed and programmed the G.E. post Government employed in the Marine Sciences branch of processor for the CIM -X tool changer equipment now in the Oceanographic Research Department. He joined use at RCA Limited. In the Fall of 1972, Dan joined RCA RCA Limited in 1970 as an Administrator of Engineering Limited as a senior NC Project Leader within the The Servo -Mechanism Laboratory at Systems within the Management Information Systems Fabrication Model Shop. department. His main function includes technical sup- M.I.T. was commissioned to work out port for the Government and Commercial Systems the details of Parson's theory. M.I.T. Division on all aspects of computer applications. Reprint RE- 20 -1 -19 developed a machine control system in Final manuscript received January 15, 1974. which the control instructions were em- bodied in a paper tape. The tape was punched by the manufacturing engineer to achieve the exact machine movements he wanted. The tape reader and servo - controls caused the machine to execute those instructions repeatedly and depen- dably. Since then NC revolutionized the metal -working art. Today, one quarter of all metal- cutting machine tools are numerically controlled and applications outside the metal- cutting field are occurring with increased frequency.

Present machines in use are a CIM -X series 20 (Fig. 1) featuring a 24 -tool drum with a 3 -axis contouring capability, and a jig borer (not shown) having an accuracy of 80- millionths of an inch.

Computer interface

Computerized numerical control applies computer languages and systems to the

www.americanradiohistory.com paper tapes to be Jsed in used in the numerical control applications at RCA Ltd. Fig. 2 - Part programs are prepared by perforating Fig. 1 Photo of a CIM -X tool changer machine - controlling part production

using a preparation of control instructions for description of the cutter tool. an operator types a part program guiding machine tools in their production 30- character; s terminal. Upon comple- a Mylaro tape will be of machine parts. Automatically A specific part program is used as an tion. punched trial part. programmed tools (APT) is the pro- input for the APT general processor. generated for production of the gramming system which generates these APT processes the input and produces a instructions. The APT process is cutter -location tape and a listing of the divided into three integral steps: I) processed part program, including Present and future usage at vocabulary, 2) part program, and 3) post diagnostics. The cutter -location tape file RCA processors. contains successive cutter location coordinates and control information concerned with placing a tool. General The vital link between the computer and APT vocabulary processor computations are valid for a the NC machine is the part program. The wide range of machine tools and their development of a part manuscript can be permits The APT programming language controllers. divided into three basic steps: geometrical the user to describe the geometric description, tool motions, and machine properties of machine parts in simple functions. terms. The vocabulary (consisting of over APT post processors 300 words) describes the various motions The first operation identifies the part for of the machine tool. The large number of APT general processor output serves as the computer. Such identification con- words allows the user to define a input to an APT post processor which sists of geometrical equations describing particular part and machine in an optimal converts the generalized coordinate the physical size and shape of the object to number of instructions. As a result, a values into a form compatible with a be machined. Dimensions are taken from large number of NC machine tool particular machine tool. Standard in- the drawing specification and formulated motions is necessary to produce a com- structions must be interpreted and into statements such as line, point, circle in plex part and can be programmed translated into correct codes that apply to and plane definitions. relatively few APT statements. a particular controller. The post processor generates a punched tape The second step outlines how the tool (Mylar® or paper) or a direct numerical (punch. drill or end mill) should be part programs APT control (DNC) file. The DNC applies to located relative to the geometrical specific families of machine tools and definitions previously stated. The tool is A systematic patterned application of the their controllers. The post processor out- driven along these surfaces or lines to APT language results in an APT part put tape, or DNC file, is then fed into the simulate the actual machine process. The program. Such a program consists of a controller to produce the desired part. choice of these motions is of great series of statements in symbolic and significance in producing an economical numerical form, representing the 2) part. The motion efficiency will geometry of a particular tool and a At the computer interface facility (Fig.

www.americanradiohistory.com PART SPELI FICATIONS

GEOMETRICAL REPRESENTATION

TOOL MOTIONS DEFINITION

DEFINITION OF MACHINE FUNCTIONS

MANUSCRIPT INPUT TO COMPUTER

Fig. 4 - The speed -brake manifold above was precision machined by use of RCA Ltd's numerically controlled PROGRAM machine processes. EXECUTION y MANUAL PUNCHED TAPE PLOT VERIFICATION most suited for numerical ,;ontrol processes. By devising a part program first, later modifications (such as design

MOUNT TAPE ON improvements and material changes) can N/C MACHINE be easily and quickly implemented. The high accuracy of NC machines is an asset in TRIAL PART the fabrication of spacecraft com- DESIGN ponents. Also in a manual system, such MODIFICATIONS design would require weeks to fabricate. The steps involved in designing and

SHOP producing a part from original FINAL PART MODIFICATIONS specications are shown in Fig. 3.

Fig. 4 illustrates a typical example of the Fig. 3 - The three basic part production modes (part program, computer interface and optimization) workmanship achieved with the CIM -X are shown here in step -by -step fashion. tool changer in machining a speed -brake manifold from a solid block of aluminum.

determine machining time. Routines The execution is performed in a time- have been developed to optimize cutter sharing mode. This type of environment Current planning travel and programming time. permits a program to be debugged and modified without the delays in tur- Plans are currently underway to incor- The final step is to include in the program naround encountered in a batch system. porate a computer -aided design and an machine functions or commands. Such automated drafting system within the functions are unique for each machine A machine tape may be requested once engineering activity. The interface type. For example, a punch press must be the execution is satisfactory; this paper between the NC shop and the new instructed when to punch a hole, when to tape output is mounted on the NC engineering facility will provide a more change a tool and which tool to use next. machine for a trial run. Optimization can effective method for the fabrication of The machine function step is of primary then be performed from this part for parts such as printed circuit boards, importance, as it will govern, on most production use. For complex parts, a plot currently one of our major products. machines, the rate at which the material of the program output can be produced will be processed. Efficient feeds and for detailed verification prior to produc- From a series of specialized compute' speeds will increase productivity and tool ing the final machine -coded tape. Flex- instructions, a drafting machine can be life. The rate of material removal (feed) ibility, the primary aim of the machine controlled to outline the layout of can also determine the surface finish of function procedure, allows changes to be printed- circuit boards. Such instructions, the part and the precision. incorporated and easily recycled in the with the aid of particular computer process. It is this adaptability feature that software, can be used to produce NC A complete part program is typed on a is of prime importance to a high- variety, tapes. Resulting from this technique a data terminal to produce a source docu- low- volume shop, prevailing at RCA parts geometry data bank could be set up ment on paper tape. The tape is then Limited. to interphase with NC programming and transmitted to a computer for processing greatly enhance productivity within the by the APT System. Prototypes and experimental models are NC shop.

www.americanradiohistory.com Data acquisition, control TODAY'S COMPLEX Naval weapons systems must be able to monitor system operability, observe performance in real and display for Navy ORTS time, and assess mission effectiveness. Having these resources, alternate T. Taylor, Jr. M. LeVarn weapon- system configurations and maintenance actions can be initiated to remedy system failure immediately. Data acquisition. control and display functions for the Navy's AEGIS Operational Readiness Test System (ORTS) is described herein. ORTS provides on -line, real -time In a multi- radar, command -and -control monitoring of the status of the AEGIS Naval Weapon System and is designed to: (a) complex such as Aegis, system perfor- determine operational readiness at system and equipment level, (b) evaluate performance mance and effectiveness criteria are ac- degradation and recommend reconfiguration, and (c) provide centralization of quired from observing the following maintenance functions for display and control which can be operated with minimum of diverse parameters and sources: personnel training. Specific goals achieved are fault detection coverage of 90- 95 ° /c: false alarm probability not greater than 10/0. This paper describes advanced techniques used in I ) normal system operation through the data the development of the data acquisition, control and display systems of ORTS applied channels and displays; specifically to the AEGIS AN /SPY -1 phased array radar system. 2) system response to periodic tests interleaved during normal mission activity; 3) response to dynamic target tests at specified on -line intervals; and, 4) results of on -line, continuously -scanned, test -point data from equipment locations. Theodore Taylor, Jr. Senior Engineering Scientist, Marc F. LeVarn - Senior Project Member, Aerospace Aerospace Systems Division, Government and Commer- Systems Division, Government and Commercial cial Systems, Burlington, Mass. received an (SB) ME Systems, Burlington, Mass., received the BE in Electrical On Aegis, these four functions are per- degree with sub -major in electronics from MIT in 1953. Engineering from Villanova University in 1958 and the formed by the MK -545, Operational He has pursued graduate studies at the University of MS in Electrical Engineering from the University of Pennsylvania and Northeastern University. He was Pennsylvania in 1962. His recent assignments have been Readiness Test Systems (ORTS). responsible for system analysis, integration, as principal engineer for concept development and diagnostics, and test planning for the USAF/USMC TIPI system design of the data acquisition and display portion DC /SR program and studies for application of a large - of AEGIS MK -545 Operational Readiness Test System Of the sources identified above, item 4 is scale information processing system to Armed Forces (ORTS) - and as a system engineer responsible for particularly important with respect to Entrance Examination Systems. Mr. Taylor was formerly preparation, review, coordination, and negotiation of impact on hardware design, since care responsible for concept development and engineering specifications and test procedures as well as integration design of the data acquisition display for the AEGIS of portions of the USAF TIPI DC /SR segment. He has must be exercised to insure that data Shipboard Operational Readiness 'est System (ORTS). participated in the design of a number of general and acquisition does not impair system per- From 1967 through 1970, he managed design and special purpose test systems including Multi- System development of time -shared computer controlled test Test Equipment (MTE), Depot Installed Maintenance formance. In this regard, item 4 is the systems for manufacturing test of cisc storage units. He Test Equipment (DIMATE), special test equipment for main subject of this paper which ad- also directed test programs related to improving the the Lunar Excursion Module Attitude and Translation acquisition and dissemination of manufacturing dresses the data acquisition, control and process Control Assembly (LEM /ATCA) and others. He is information. From 1962 through 1967, Mr. Taylor was currently studying the transition of the AEGIS ORTS display elements of 'ORTS. responsible for development of automatic test systems system from the engineering development model to a for the Apollo Lunar Module. His work involved mission modified version for the DG class of ships. Mr. LeVarn is System requirements simulation, automation of stimulus and measurement. a member of IEEE and a registered engineer in the State and probabilistic models for checkout effectiveness and of Massachusetts. confidence levels, for S -band communications, X -band radar and transponder, and attitude propulsion control The basic system requirements for ORTS systems. Mr. Taylor has 20 years experience in develop- Reprint RE- 20 -1 -23 were derived from functional and test ing electronic systems for DOD and NASA and has Final manuscript received Jan. 2, 1974. requirement analyses the published numerous papers in this field. He is a member dictated by Since this article was written, Mr. of IEEE and AES professional group; he is a registered Taylor has left RCA. Overall Aegis Mission. Aegis (Shie'.d of professional engineer in the State of Massachusetts. the Fleet) is a total weapon system designed to provide wavetop -to- stratosphere area defense of the fleet. It provides total integration of target detection. command and decision, and weapons deployment within the Aegis ship and the total escort force. The heart of the system is the multi function array radar (AN /SPY -I) which detects and tracks multiple targets at all ranges.' Illumination radars provide missile control and target illumination for guidance, homing, and kill evaluation. Overall control and display of the weapon system environment is facilitated by a complement of AN /UYK -7 computers and AN /UYA -4 display consoles located in the combat information center (CIC).

The functional requirements necessary to

www.americanradiohistory.com closely integrated into the weapon ESTABLISH AND RESPOND TO system. This was a logical conclusion in a) MODE MONITOR RECOMM ND 6) SCHEDULE STATUS ON EQUIPMENT that over half the operational- readiness RECONFIGURATION c) CONSTRAINTS LINE test information is derived from the prime 1 equipment's normal operating data DIAGNOSE AND IDENTIFY IDENTIFY channels, utilizing ORTS computer - FAULTS AND EVALUATE DETECT SYSTEM controlled simulation tests for mode READINESS -e OPERABILITY CONDITION LEVEL The ICON DUCT ON -L INE operability and target track. OPERABILITY TESTING remainder of the test data and much of CONDUCT CORRECTIVE fault- isolation information are obtained PERIODIC MAINTENANCE SYSTEM through test -point data sources spread TESTING throughout the system and in remote ship locations. SUPPORT TRAINING The EDM -I / MK -545 ORTS configuration and its interface with the Aegis weapon system are shown in Fig. 2. Fig. 1 - ORTS functional flow diagram. Central control of ORTS is provided by the Test and Monitor (T &M) console. support Aegis dictated that ORTS AEGIS are: Tests are scheduled, controlled, and provide: reported on automatically by program 1) Fault detection coverage (90 -95 %) control from the computers, with access a) Determination of operational readiness at 2) Monitoring accuracy (

naC,C., I AMOY

; SYSTes r `-1

rwrs.""""^"" I r ---- .4 MIN COMNITII 5ia,u.,x.so. COI rtar. r-' l

I nuaur )ra.rn I L- _Q- ö:,..,C J _____ L___J

r I

a... I ..a...... ---J

111,IIIII WIT I

i .....s ` "Y 1 _ J

Fig. 2 - AEGIS /ORTS, EDM -I /MK -545 configuration

www.americanradiohistory.com Retrieval of remote test -point data and the role of the T &M console are major elements of the system. Acquisition of remote test -point data is essential not only for failure detection, but as the most expedient means for reliably isolating failures to repairable levels. DATA ACQUISITION The test -point data is gathered by means ASSEMBLIES of data 'acquisition assemblies (DAA) resident in each weapon system cabinet. Each DAA is serviced by a common data TEST AND MONITOR AND MONITOR CONSOLE bus from the T &M console. In response STATUS CABINET to sequentially received address and clock data, the DAA's transmit measured test - SIMULATORS COMPUTER point data back up the bus. As many as PR EC ONDITIONERS PROGRAMS sixteen buses with a maximum of 512 test - point addresses per bus may be used. The sequence of addresses on the bus as well as the processing of received test data are computer controlled through the T &M console.

Critical to the design of the ORTS system Fig. 3 - Placement of ORTS equipment aboard ship. was the determination of data bus characteristics, data rates, methods of coding, cable immunity to noise and EMI, cable length, test -point precon- Transmission system Design analyses ditioning and fail sate provisions. A discussion of analysis, tradeoffs and Two basic methods of data acquisition In arriving at the data -bus design, several design details of these areas follows. were considered. The multiplex method additional factors and tradeoffs were (Fig. 4) employs sequential switching of considered: remote test points to a central set of signal Data gathering conditioning equipment where the data is Test requirements analysis converted to a standard data processing format. The master multiplex -sub- Binary coded transmission vs signal gated A test requirements analysis (TRA), multiplex method is a variation using two transmission. (initiated during contract definition or more levels of multiplexing (e.g. a gated transmission of data was and later expanded during engineering master multiplexer within the equipment Signal considered because of its immunity to development.) identified significant room with a sub -multiplexer within each single bit error. However, it would have test parameter and measurement crite- cabinet). Multiplex methods ordinarily required either much higher data rates, or ria. The analysis encompassed the fol- acquire test -point data at a fixed se- greatly extended data transmission times lowing: quence and rate, and are well suited for to obtain analog resolution equivalent to a) Test parameter purpose (failure detection or monitoring continuously available, slow- of a I0 -bit binary code. Also, this fault isolation; ly varying analog signals. A major short- that technique would have been incompatible b) Effect of system operating mode on test coming of the multiplex system is the with the data transmitted from digital test point function; impact of a failure at the multiplexer c) Nominal value, tolerance and measurement control or the central signal conditioning. limits; d) Noise, ripple and common -mode rejection; The data bus (Fig. 4) uses individual e) Pre -conditioner and stimulus requirements; signal conditioners for each test point. Moreover, each test point is individually f) Physical location. .co,,. -- addressable. The use of individual signal A diagnostic flow chart for each major conditioners has been made possible by equipment illustrated the logical progres- the advent of large scale integration sion of the diagnostic process from fault (LSI). This same technology allows in- detection to isolation of the failed corporation of register storage, making it replacable sub -assembly or sub -assembly possible to acquire high- speed -digital group. data asynchronously to weapon system A significant result of the TRA was the timing. The data -bus approach reduces determination that all test -point data the impact of circuit failure. and could be preconditioned to one of three minimizes the required cabling. Sequence standard dc analog 0.25 to of test -point acquisition is outputs: completely Fig. 4 - Data acquisition approaches showing the single 6.67V, serial digital, and parallel digital. under computer control. multiplex and multiple signal conditioner arrangement.

www.americanradiohistory.com 4 7 9 101JI1IJ12IJ13 ANALOG, CLOCK J tJ 2 tJ 3 IJ5 LJ6 LJ IJ SERIAL DIG. 2II-APSEIC AND PARALLEL -I

II I 1 II 11 I DIGITAL II I II y11 TEST PARTS r 0 0 t 1 t 1 1 D t D t 1 D t 1 t 0 A t

ADDRESS G TEST POINT P

A I S ADDRESS (TP229) R i

P COMPUTER R Y VO CLOCK (CONTINUED) 14 15 16 17 18 19 20 U 21 i--I 22 tJ 23 U 24 U 25 Q +0+0+0+0+ 1+ 1+0 +0 +1 +ly 1

M REPLY MEASURED VALUE Pe N (ANALOG REPLY 5.37 V) AI RT

Fig. 5 - The ORTS data acquisition system. Fig. 6 - Data bus transfer timing.

points and would have complicated the the parity approach was adopted. Also to 20 DAA's may be "daisy chained ". processing circuitry in the T &M console. test point addresses which differ by only Each data bus has three twisted shielded one bit are avoided to preclude noise - pairs with an overall shield and armor Although the impact of a single bit error induced addressing errors. where requir &d. Two of the pairs are used due to random noise is much greater in to transfer clock and address signals, to the binary system, it was chosen because Data rate the DAA's and the third pair carries data the probability of error was calculated at reply signals from the DAA's to the T &M less than I pai t in 1010. As an additional A relatively conservative clock rate of 20 console. precaution, the software was so designed KHz was chosen. This permits acquisi- That all out -of- tolerance values are re- tion of approximately 770 test points /s Interrogation of a test point consists of sampled for confirmation before any on each data bus, and 6150 test points /s transmitting 10 bits plus parity of address failure indication is displayed. with all buses operating in parallel. It is data from the T &M console to all DAA's more than adequate to meet the perfor- on a data bus. The particular DAA which Biphase vs non -return -to -zero (NRZ), or mance monitoring needs determined by recognizes this address responds with 10 pulsewidth coding the test requirement analysis. bits plus parity of measured value. A biphase- coding scheme was adopted Twenty-five 20 -KHz clock pulses are because the maximum pulse width is only The 20 -KHz rate also provided very wide transmitted from the T &M console as one cycle of the clock frequency. latitude in choice of technology for the timing references for this data transfer Therefore, ac coupling is used at the DAA design. The long multi- tapped bus (Fig. 6). Both the address and data signals drivers both in the DAA and the T &M structure requires use of conservative are transmitted in biphase code. The console; thus, dc ground loops are avoid- data rates, and noise immunity is en- address bits are weighted in binary ed, providing a common -mode immunity hanced by the use of a restricted fashion; the reply bit weighting depends that contributes to fail -safe driver design. bandwidth system. upon the characteristic of the parameter Ac coupling would not be possible with being monitored. NRZ or pulse -width transmission techniques. Data transmission characteristics DAA characteristics Parity vs address retransmission The final transmission system design uses The TRA identified three types of signals A positive safeguard is required to avoid a data -bus approach with limited mul- most commonly existing at test - point acquiring data from an erroneous test tiplexing, providing the advantages of the interfaces: analog, serial digital, and point due to circuit failure or noise data bus at a reasonable cost per test parallel digital. Standardization of three induced on the data bus. Use of a parity point (Fig. 5 illustrates this system). Each monitoring interfaces permitted develop- bit was considered, as was an address prime equipment cabinet contains all ment of a minimum number of new retransmission scheme (receiver returns multiplexing, test point selection cir- custom integrated circuits and hybrids to sender the address received). The cuitry, and common -signal conditioning for use in the DAA, while not creating chosen parity approach minimizes circuit in one data acquisition assembly (DAA). unreasonable preconditioning problems complexity and data acquisition time, Only signal conditioning unique to each in the prime equipment. while providing protection against single - test point is exterior to the DAA. Eight bit errors equivalent to address data buses extend from the Weapon The DAA (block diagram Fig. 7, retransmission. Address retransmission System Test and Monitor (T &M) console photograph Fig. 8) contains a quantity of offers superior protection against two -bit into the AEGIS equipment rooms. test acquisition modules (TAM's) of three errors; however, the probability of two - types corresponding to the three monitor- bit errors was found to be negligible, and Each bus may be up to 400 ft long, and up ing interfaces identified above. A single

www.americanradiohistory.com DAA contains slots to accommodate up 2) its analog conversion accuracy was indepen- to 13 TAM's. Nine slots are capable of dent of input -clock accuracy, and accepting either analog or serial digital 3) the critical parameters for assuring satisfac- TAM's. The remaining four slots are tory performance were more easily identified and controlled for this approach. capable of accepting analog, serial digital or parallel digital TA M's. Serial digital monitoring

Analog monitoring Each serial digital TAM is capable of Fig. 8 - Data acquisition assembly with cover removed showing modules placement. monitoring 16 test points. It does this in Each analog TAM is capable of monitor- conjunction with local storage cards ing 16 differential inputs. Of these, 15 are located within the prime equipment. used for prime equipment test points, and way as an analog test point. The results of one connected to a calibration standard this measurement are used by ORTS Whenever any one of the 16 addresses internal to the DAA. Each of the 15 software to perform offset calibration on associated with the TAM is recognized, analog test points monitors a voltage in all other measurements from that TAM. timing is initiated, resulting in a burst of the range from +0.25Vdc to +6.67Vdc; This permits ±30- millivolt accuracy shift pulses on the shift -pulse output line. the range was established by MIL -S -1326 is to all 16 local measured at the ideal switching points of This line "daisy chained" which was used for design guidance. the TAM analog -to- digital converter. storage registers associated with that TAM. One of these registers is enabled by Associated with each test point is a the setting of the appropriate "address unique test -point address established by Analog -to- digital converter recognition output" line (Fig. 7). The hardwired interconnections internal to design shift pulses transfer the data stored in the DAA, but external to the TAM; thus, local storage to the TAM which generates all TAM's of a given type are identical. A tradeoff study compared the various parity and transfers data plus parity to Each TA M compares all addresses on the alternative analog -to- digital conversion the data bus. address bus to its set of wired -in ad- techniques for use in the analog TAM. dresses and responds when comparison is Included were: voltage controlled os- successful. Upon receipt and recognition cillator, single ramp, up/ down ramp, and Parallel digital monitoring of a test -point address, the analog TAM successive approximation. Only the dou- converts the test -point input to a serial 10- ble ramp and the successive - Each parallel digital TAM is able to bit plus parity binary representation. All approximation techniques appeared monitor four groups of 10 bits of parallel address and timing signals are received on capable of meeting the accuracy re- data. Each group may represent data common address and clock buses. The quirements. Successive approximation from one or more parallel digital test serial binary output is transmitted on a was selected because: points. A strobe signal may be used to common reply bus. load 10 bits of information into the TAM, or the strobe may be wired I) it naturally generated a binary output which permanently The internal calibration standard has its was most desirable for data -bus transmis- to a strobe -enable line. When the strobe is own address and is measured in the same sion, permanently enabled, the local storage will continually reflect the state of the data inputs.

TP INPUTS' TO OTHER TAMS Whenever one of the four addresses ANALOG IN DAA 14 TAM associated with the TAM is recognized, it CALIBRATION REFERENCE will shift the associated 10 -bit group plus (PORTION OF (5 DAA POWER parity in serial onto the DAA common SUPPLY)

o , reply bus. This output is non -destructive,

I 10 I and the data remains stored in the TAM TP I INPLTTS I 7 for possible re- interrogation. o u` SERIAL 5 DIGITAL TAM o ADDRESS o. Each parallel digital TAM provides four RECOGNITION 15 OUTPUTS 0 address recognition outputs. These lines SHIFT PULSES a 6 GRD LINES R may be used as control inputs to the prime equipment to be available to initiate test 0 1010 DATÁ TPO TR operations.

o RD a o CLOCK TPI PARALLEL DIGITAL TTL compatibility

44. TAM ADDRESS REPLY O m V TP2 All outputs from both the serial digital and parallel digital TAM's to local e e o e c TP3 preconditioning are TTL compatible. All TO DATA BUS inputs to these TAM's from local preconditioning are TTL compatible, provided Fig. 7 - DAA block diagram. external pull -up resistors are added to the

www.americanradiohistory.com TTL output stages in the local precon- Status displays iitioning. Status display panels provide indications TAM design of weapon system function status, and mode availability. Sufficient detail per- All Tam's are combinations of PMOS mits the maintenance supervisor to deter- arrays and hybrid devices mounted on mine the presence of prime equipment multilayer printed circuit cards whose and ORTS failures, functionally locate dimensions are about 4.1 x 3.1 inches failures, and permit preliminary assess- (Fig. 9 shows top views Fig. 9 - Top views of an analog TAM (left) and serial of an analog and a digital TAM (right). ment of the impact on the weapon- system serial digital TAM). operational capabilities. It also provides initial display call -up guidance to the Various combinations of LSI and hybrid hybrid development was initially done by maintenance operator in the use of the technology were considered during the the custom hybrid facility at ASD, interactive data display for detailed fault TAM design. PMOS, COS / MOS, and Burlington. External second sources were isolation. Display functions are im- bipolar LSI using manual custom later utilized for some circuitry. plemented by using rear -projection in- designs, computer -aided designs, and dicators; thirty six such indicators display gate -universal array approaches were all the status of the major hardware func- appraised against these criteria; speed, Control and display tional elements for the radar. Twenty - power, consumption, size, reliability, four indicators are provided to display cost, and delivery. At the time of the The Weapon System Test and Monitor the availability of the radar operating tradeoff, it was decided to use PMOS- (T&M) console provides a centralized modes. Sixty indicators display func- gate universal arrays developed by RCA facility in the preventive maintenance tional status of the Fire Control and in Somerville to accomplish all logic room to operate in conjunction with Weapon Direction Segments. functions. The PMOS approach employs other ORTS equipment (and AEGIS standardized gate structures inter- computers) for monitoring and testing All function status and mode connected by a customized final metaliza- the weapon system; it provides a mul- availability indicators a tion layer. tilevel display of operability status and employ color code to indicate the status of the function mode as performs the following functions for the or This provided LSI circuitry at a EDM -I AEGIS system: follows: significantly lower non -recurring cost 1)0K (Green) and promised faster initial delivery than a) Provides summary status of the Radar. Fire other approaches considered. The PMOS Control, and Weapon Direction Segments 2) FA ILED (Red) of circuitry offered adequate speed and AEGIS. 3) DEGRADED/ BACK-UP (Yellow) power consumption well within the b) Provides I/O communication with two 4) OFF (White) computers. ORTS constraints. The PMOS 5) NOT MONITORED (indicative of ORTS technology was also known to be capable c)Controls acquisition of sensor data. failure -white with red diagonal stripe). of performing satisfactorily under the d) Provides a 2 -way data link to deckhouse Summary displays of maintenance mode environmental conditions of shipboard monitor and status cabinet. e) Provides a and mission status are also provided. use. COS/ MOS computer -aided design capability to initiate special test and monitoring routines (through operator would be more seriously considered to- interfaces). day because of technology ad- Data display and keyboard vancements. The operation of the (EDM -I) T &M console (Fig. 3) is shown by the block The console contains a data display unit It proved possible to standardize two diagram of Fig. 10. The console corn - employing a CRT display detailed address and control arrays which per- municates with two AEGIS computers. of form the functions of interfacing the Ordinarily, the radar computer controls maintenance information. The CRT dis- lines TAM's to the two DAA data buses, all devices on the FWD-I/0 bus, and the plays 38 of 80 alphanumeric address comparison, test -point selection, weapon direction (WD) computer con- character spaces each. The upper and and of parity-checking- and -generation. trols all devices on the AFT -I /O bus. lower two lines are protected from These arrays are identical on the serial During reconfigured operating modes, keyboard data entry or erasure, and are and parallel digital TAM's, and are iden- either computer may access any device. used by the two AEGIS computers inter- tical in design to the equivalent arrays on The switching required is accomplished facing with this console. Thus, the analog TAM, but are fabricated using within the I/O routing circuitry. information pending may be called up for a different process to achieve operation at display by use of the keyboard. The different voltage levels. The external interrupt electronics detects, summary- status display panels previous- stores, and encodes keyboard parity ly described also provide guidance in the All precision analog signal switching and error, time out and power on master reset call -up of detailed displays. Display Gall- conditioning required for analog TA M's (PM R) interrupts. The interrupt elec- up is accomplished by typing a code on is performed within a hybrid circuit. tronics establishes the priority for each the keyboard and entering it into the Other hybrids are used to provide the interrupt and initiates the transfer of an appropriate computer, using one of two high -drive levels required by the digital interrupt code, based on its priority, to "enter" keys on the console. The display - TAM outputs which drive TTL. Voltage the computer designated by the interrupt and- control characters used in operation regulation is also done by hybrids. All code. of the data display unit and keyboard are

www.americanradiohistory.com those of the ASCII code set. sub -assembly groups. arrangement serves as a remote extension d) Output of any one test point periodically of the T &M console forward and aft I/O The data display unit is capable of updated. buses. Display information is outputted accepting data from either the forward or to the monitor and status cabinets and aft console I/ O bus, but only from one at Teleprinter requests for display generation and a time. It will output a "busy" status word special test initiation are received from if the second bus attempts to access it these cabinets. A tradeoff was performed The T &M console contains a teleprinter before the first bus terminates. to ascertain the optimum method for data providing a capability of obtaining hard transfer. It was concluded serial data copy output of any display on the data that Data on the keyboard is transfer using parallel control and in- entered dis- display unit. This is accomplished by terrupt lines reduced the number of cable played on the CRT in the position in- depressing a print key on the keyboard a cursor. The initial conductors greatly over parallel data dicated by cursor causing the contents of the data display is transfer. position controllable by the computer, unit memory buffer, including necessary or by the operator using special keys on control codes, to be outputted to the the keyboard. The size and position of the All data and continuous control signals teleprinter. The minimum teleprinter screen portion protected from keyboard are transmitted in biphase form, permit- print speed is 50 lines per minute. data entry or erasure are variable under ting ac transmitter coupling and the computer control. An audible alarm is resultant high common -mode immunity. provided and sounded upon receipt of the The data transfer rate is 150 KHz; the ASCII bell code. Keyboard editing con- Data acquisition electronics similarity in data transmission technique trols (in addition to those for cursor between the deckhouse interface and the positioning) include screen erase, line The T &M console includes circuitry for data acquisition bus interface permits a erase, and cursor return which ac- routing data between the radar corn- common design for the T &M console complishes a partial line erase. puters, and the data acquisition buses drivers and receivers for these interfaces described earlier. Each bus is controlled even though the data rates differ. by an acquisition control module (ACM). Display formats The ACM's in turn are controlled by a In addition to the signal interface test control module (TCM). The TCM between the T &M console and the The display format and content is com- controls eight ACM's and is designed to deckhouse monitor and status cabinets, a pletely under software control, and is permit all ACM's and their associated remote power control interface is also subject to continuing evaluation and up- data buses to operate in parallel. provided on the T &M console. This dating. The following are among the makes it possible to apply power to a display possible associated with test point monitor and status cabinet, provided that data acquisition: Deckhouse interfaces power is available in the equipment room. Indications of power availability and a) Results of periodic monitoring scan of application are provided functionally related test points with indica- The T&M console contains deckhouse on the T &M tion of measurements out of limits. control circuitry which permits bi- console. b) Results of diagnostic measurement se- directional communications between quence with indication of measurements out console, and the ORTS deckhouse of limits. monitor and status cabinet located up to Deckhouse monitor and c) Identification of failed sub -assemblies or 400 ft. away from the console. The status cabinet

The deckhouse monitor and status DATA ACQUISITION BUSES TO EQUIPMENT ROOMS cabinet provides a local display and control capability in the AEGIS equipment rooms. It contains summary t t status displays similar to those on the T &M ACM -F 2 I ACM -F3 - __- __IACM -FB console. It does not contain a CRT, but TO does employ a printer to DECKHOUSE output detailed I3 DECKHOUSE RADAR MONITOR d CONTROL CONTROL diagnostic information. A keyboard is STATUS STATUS TCM ELECTRONICS ELECTRONICS DISPLAY CABINET used to input data requests and control information. The cabinet also contains a TO D I/O BUS RADAR AE---. IF 14-character alphanumeric message dis- COMPUTER play, using light- emitting diodes. The message display is used to compose and I/O EXTERNAL CONTROL DATA -- INTERRUPT TELEPRINTER ROUTING ELECTRONICS ." DISPLAY edit keyboard data messages before ELECTRONICS UNIT processing them in a computer. It is also TO WD used to view test -point measured values, 7-AFT VO BUS COMPUTER - as requested from the keyboard.

FCS/WDS CONTROL DATA STATUS ELECTRONICS ACQUISITION Reference - DISPLAY ASSEMBLY I. Bernstein. F., and Strip, J., "Aegis Engineering Model Fig. 10 - Weapon system test and monitor console block Command and Control System ", RCA Engineer, Vol. 19, diagram. No. 5, Feb: March 1974.

www.americanradiohistory.com Low -cost distributed -port reflex loudspeaker enclosure C.C. Rearick

The author discusses the design of a loudspeaker enclosure, cost competetive with existing products on the market. The loudspeaker system is optimized for the best possible sound response consistent with cost. To reduce distortion at low frequencies, the reflex concept was incorporated. The paper describes the design restrictions, theoretical restrictions, design analysis, actual results, and conclusions. Illustrations include frequency response data, impedance characteristics, and distortion data.

SMALL LOUDSPEAKER SYS- output power, requiring that voice -coil TEMS have not commonly been displacement vary inversely as the square designed using the reflex -enclosure ap- of the frequency -and increase still proach. Instead, it has been easier to use further near resonance. A loudspeaker the acoustical- suspension approach capable of such large linear displacement, wherein the compression of the air inside especially in the 5- inch -size range, the sealed enclosure provides the restor- becomes inordinately expensive coming force or compliance element of the pared to the reflex enclosure scheme speaker suspension system. where the realizable benefits are also greater. Reflex versus acoustical suspension The reflex enclosure, in contrast with the acoustical- suspension approach, can be An acoustical suspension system is designed to acoustically load the rear of almost foolproof from a design stand- the loudspeaker within a wide range of point, but the loudspeaker requires an frequencies, and thereby contain the unusually wide -displacement capability. voice -coil motion to reasonable ex- The wide displacement need results from cursions. The bulk of the sound energy, the mechanical relationship for constant instead of radiating from the speaker cone, now radiates from the enclosure Fig. 1 - The author is shown with the model 12R -410 distributed port, reflex loudspeaker enclosure. Both the vent or port which is not hampered with prototype and final product are pictured. restrictions in displacement as is the acoustically suspended speaker. Since large excursions of the speaker are not needed in a reflex -enclosure design, a

Charles C. Rearick, Engineering, New Products, Parts significant saving in loudspeaker costs and Accessories, Deptford, N.J., joined RCA in 1956, results. having graduated that year from the University of Maine with the BSEE. He completed the engineering training program and then served two years of military service as an army lieutenant. He returned to the Home Instrument Reflex enclosure principles Division engineering labs at Cherry Hill, where he became responsible for the acoustical and audio performance of the TV product line. After one year with the An open-ended tube, when inserted into M&SR Division, he transferred to the Camden plant where he participated in various audio, digital signal the side of a closed box, effectively shunts processing and magnetic recording projects. He the acoustical compliance of the box with received his MS in Engineering for Graduate Work in an acoustic mass equivalent to the mass Electrical Engineering from the University of Penn- sylvania in 1969. As Technical Publications of the air in the tube. The result is Administrator for Parts and Accessories, Mr. Rearick is analogous to placing a parallel LC reso- also responsible for the review and approval of technical papers; for coordinating the technical reporting nant circuit in series with the mechanical program; and for promoting the preparation of papers for the RCA Engineer and other journals, both internal and impedance network of the loudspeaker. external. This resonance can be effectively utilized in a loudspeaker system by tuning the box resonance to about one -half octave below Reprint RE- 20 -1 -15 Final manuscript received May 3, 1974. the resonance of the loudspeaker. It is in

www.americanradiohistory.com ELECTRICAL --\ ,r-- MECHANICAL ,t-- ACOUSTICAL- RIM SUSPENSION, CHI Z BI :I SDI SPKR. - ---i RADIATION CONE,MM ,and AREA, SD Z mst i I 7 C I;T VOICE COIL SUSPENSION, CM 2 Blv f=BII ( v) P VOICE COIL MM PORT 2 f1 RADIATION

FLUX DENSITY, B GAP VOICE SPKR. BOX PORT WIRE LENGTH, I COIL

Fig. 2b - For loudspeaker design and analysis, the circuit above has an electrical analog (at left), a mechanical analog (center). and an acoustical analog (at right). The three equivalent circuits are coupled by Ideal transducers.

Fig. 2a - Sketch of the loudspeaker component parts.

this region where the linearity of the been designed (Fig. I) at RCA Parts and trodynamic loudspeaker whose voice coil loudspeaker suspension is sorely strained Accessories according to this premise, exhibits the usual characteristics of and represents the principal source of non and is represented in our product line by resistance and inductance. Our first act of linear distortion in most loudspeaker model I2R -410. It was designed to com- simplification will be to ignore the induc- systems. From an octave or so above plement the RCA car stereo -tape player tance since it is insignificant at the low resonance, the mass of the diaphram and line for use in the home with the model frequencies we are considering (30 to 1000 voice coil limits the motion well within I2R -900 and I2R -1000 car stereo -tape Hz). the linearity constraints of the suspen- player, home -converter models. sion; below this point, the motion becomes progressively more compliance- The I2R -410 utilizes a driver costing Therefore, our first approximation of the controlled. The restriction of voice -coil considerably less than acoustical- system consists of an electrical resistance motion caused by the loading of the suspension types of drivers -and is called R,r. The ideal loudspeaker (Fig. 2a) resonant box on the rear of the patterned after the very successful driver will move whenever current flows diaphragm will obviously result in used in the model 12R -400 car stereo through its voice coil. As soon as we get reduced acoustical output from the speaker installation kit. This speaker, motion, in the mechanical circuit (Fig. speaker. However, the volume velocity of however, has especially tight 2b), we enter the realm of the mechanical the sound energy (acoustically analogous specifications on its fundamental world wherein the system contains a mass to current) circulating in the resonant resonance, a cloth -rim suspension, and that must be moved, and a compliance.or box, radiates a sound pressure from the an oversized voice -coil suspension. The spring element to be stretched. port almost totally unrestricted by non- system performs very well down to fre- linear suspension elements. Such radiated quencies nearly as low as those reached At some frequency (in loudspeakers, a sound is of sufficient magnitude and by an equivalent sized acoustical suspen- low frequency is most expedient), the phase to augment nicely any loss of the sion system, exhibits better damping of mass and compliance are in resonance. sound from the loudspeaker diaphragm. transient signals and performs as well or Since a mechanical system such as this better with respect to non -linear distor- exhibits a large displacement at Unfortunately, the Q of a reflex enclosure tion. resonance, electrical engineers will envi- is generally quite high so that the sound - sion immediately an electrical analogue energy- conversion efficiency in the vicini- Analysis of pertinent of a series resonant circuit (Fig. 3), and ty of box resonance is also high. This technology will include enough mechanical resistance effect is responsible for the juke-box-bass to be realistic. so typical of the many improperly The following is a review of the design designed bass -reflex enclosures one hears procedure, possible alternatives, and a all too often. By introducing a controlled discussion of the performance of the final Acoustical properties amount of acoustical resistance in series configuration. The understanding of of a closed box with the port mass, the Qcan be reduced loudspeaker enclosure design is best sufficiently to reduce the peaked response taken in incremental steps. Analytically, it in primarily and blend with the radiation from the we choose to start with the loudspeaker At low frequencies, a box acts loudspeaker itself. A distributed port, voice-coil terminals, adding in the upon the loudspeaker as a pure com- one consisting of a set of small holes various mechanical and acoustical pliance (spring load). This load combines of the instead of a single large one accomplishes elements until the solution is either ade- directly with the compliance the sound blending. quate or so complex as to defy solution. loudspeaker suspension, raising its resonant frequency accordingly Fig. 4). The acoustical compliance (CAB) of a box at The 12R -410 loudspeaker Mechanical properties of a loudspeaker low frequencies (below 500 Hz) is a direct function of box volume and is givens by A loudspeaker/ enclosure system has Figs. 2a and b show an ideal elec- the equation:

www.americanradiohistory.com ZMS Z MS S / (Sp) AR MMS M M Cm RMS CMS RMS MS R I Acoustical (SD) Q(SD1ZAR r° v C_ v AR(spkr) r°H 2 (port) Force Load Force Force velocity velocity Box Compliance velocity CAB$SD (S ) MA L D) ZAR L- (Sp o-(SD) ZAB L p Fig. 3 - Mechanical analog of the loudspeaker without Fig. 4 - Mechanical analog of loudspeaker inside a Fig. 5 - Mechanical analog of loudspeaker inside a baffle (enclosure). closed box. (ported) reflex enclosure.

C 1B= VB/PaC2= m5/ N (1) acoustical mass. The so called Acoustical mass: MAB=7.83 "distributed port" configuration, con- =kg/m4 where VB = box volume in cubic meters; sisting of a set of small diameter holes, p,= density of air = 1.18 kg/ m3; and c = offers both acoustical mass and moderate Measurements of the frequency response, velocity of sound (in air) = 344.8 m /s. resistance; and, it offers the convenient voice-coil impedance, and distortion were design advantage of being constructed at taken for the closed box enclosure - and Associated with the acoustic compliance the rate of one hole at a time until the the results are illustrated in Figs. 6 and 7. of the box is a small acoustic mass of air response is proper. The contribution of a which moves within the box as if the box set of N holes to the elements of port mass Reflex box design were a closed pipe of length (l'). This (MAP) and acoustical resistance (RAP) is acoustic mass is calculated according to calculatea )rom the following:34 Careful testing showed that about ten the relationship': 1/2 -inch diameter holes in the 3/8 -inch- Po thick front panel were about right for the MAR= l'pol3Ar = kg/ m4 (2) MAP = [P+11 N7ra loudspeaker proposed for the design. Calculation of the theoretical acoustical where A r = Cross -section area of box in m4 = kg/ mass (MAP) and acoustical resistance m2. = Acoustical mass (4) (RAP) of a set of ten 1/ 2 -inch holes (3/ fl- The complex acoustical impedance of the inch deep) from equations (4) and (5) P. (2wµ)ill'/ - closed box is then the algebraic sum of the RAP = a + (2)] results in the following: reactances of the compliance and mass N7ra2 relationships of equations (1) and (2); for example: = Acoustical ohms, mks (5) MAP= 18.93 kg/ m4 where N = number of holes; a= radius of ZAB closed box = t (1/ at CAB) + t(W MAB) RAP= 45.64 (f"'), acoustical ohms one hole; 1' = length of one hole; l" = end = mks, acoustic ohms for hole (2)(0.85)a; w (3) correction = = angular frequency = 27w/; and u= viscosi- ty coefficient = 1.56X10 -s m2 /s. where at = 27rf. Though the port is principally a mass element at this frequency, the distributed To convert the acoustical impedance of port is almost three times lower in Q than The resonant frequency of the reflex - the box, ZAB, into a mechanical im- a typical open -hole vent. The lower Q is a enclosure box is then found from the pedance, ZM,B, as seen by the loudspeaker favorable condition for reducing the classical formula for parallel tuned cir- force element, BI(i), the acoustical im- bass- boominess referred to earlier. Vent cuits, using results of equations (I), (2), pedances must always be multiplied by Q, for the distributed port at the frequen- and (4): the square of the effective area of the cy of 100 Hz, is 27.33. loudspeaKer piston, SD2. This conversion MAP MAB)( CAB)]'l2= Hz (6) is represented in Fig. 2b by the ideal fe= I/27 [( + From these results, the acoustical coupling ratio between the mechanical resonance (fb) of the box can be and acoustical networks, SD/ I. calculated to be l 1 1.2 Hz, using equation Experimental box design (6). This agrees quite closely with the As mentioned previously, an open -ended experimental results for the reflex tube inserted into the side of the box enclosure, as indicated by the minimum Closed box design effectively shunts the acoustic com- point between the two peaks on the pliance of the box with an acoustic impedance curve of Fig. 6. inductance equivalent to the mass of the A box, having a volume of 656 cubic air in the tube. This acoustical parallel inches (0.01075m) was built out of 3/8- resonant circuit appears in series with the inch -thick plywood. Computation of the Open -back enclosure loudspeaker series resonant circuit as acoustical impedance elements of this illustrated in Fig. 5. The possibilities for box from equations (1), and (2), result in The loudspeaker was also tested in a large port configuration are legion, and the following: open -back enclosure backed with a heavy authorities do not agree in all cases as to fiberglass -lined corner. Under these con- which is the best. Our problem, however Acoustical compliance: CAB = 7.66x10-8 ditions, the speaker performance is in- is specific in that we desire a lossy =m5 /N dicative of what the loudspeaker itself is

www.americanradiohistory.com response Table I - Performance specification on 12R -410 home fundamentally capable of doing (i.e., Distortion converter sound reproducer unit. without the assistance of an enclosure). The difference between the performance The loudspeaker used in the design of this System of the loudspeaker under this condition enclosure system was selected and and the performance when mounted in specified to have a free -air resonance Frequency response 60 to 14,000 Hz Sensitivity 95 dB closed or reflex boxes is il- lower than that generally either the slightly Power handling capability 3-Ii 2 W nominal lustrated in Fig. 6. The net effect the regarded as standard for its size (see Table 7 W maximum different enclosure designs have on I). The low resonance, however, can be Distortion @ 7g mW input power: speaker performance is apparent. accommodated to advantage with the @70 Hz 59ó reflex- enclosure loading technique ex- @ 100 Hz <2e4 <2% plained earlier. Fig. 7 illustrates the effect @ Second resonance the enclosure on the linearity of voice- Discussion results of Driver of coil motion. Below box resonance, (about Nominal impedance 3.2 ohms 100 the reflex Hz or so), enclosure reacts Voice -coil diameter 3/4 in. Frequency response much as an open -back enclosure, while Loudspeaker diameter 130 mm (5 -1,4 in.) the closed -box compliance characteristic Free -air resonance 120 Hz±15qí. Ceramic magnet The improvement in frequency response takes over almost completely from the Weight 130 gram (4.59 oz.) gained through the use of the reflex loudspeaker suspension compliance. Flux density 9000 gauss Between 90 and 500 Hz, however, the design is evident in Fig. 6. The improve- Enclosure hu.t dimensions ment begins at about 70 Hz, reaches a enclosure is very effective in its control of maximum at 100 Hz, and tapers off the voice -coil motion, reducing distortion Width 7-3 8 in. Height 10-3'4 in. 5 smoothly to merge with the closed box within this region by as much as dB over Depth 9-1!4 in. response at about 500 Hz. The result is a the closed -box case, and up to 3 dB over Volume 656 in.' leveling off of the drooping response in the open -baffle case. Since this is the this region. The ragged low- frequency region in which the maximum amount of reproducing unit. response of the same speaker in an open fundamental program energy is concen- back (large) baffle, shows the inability of trated (and there is actually an increase in The frequency response is smooth and this speaker to control its motion ade- sound output), the improvement is much adequate at the high end and is optimized quately through its own suspension more significant than the loss. at the low end for smoothest response and elements. The mid -range and high - low- harmonic distortion. Table I is a list frequency responses of this loudspeaker Conclusions of the performance specifications of the are excellent. The responsibility for this 12R -410 loudspeaker and enclosure. however lies with the manufacturer of the The enclosure utilizes only a single, five - speaker - and not with the design of the inch (130 mm) diameter, low -cost enclosure. Though the enclosure cannot loudspeaker in a 656- cubic -inch box with References usually improve the high- frequency a distributed port consisting of ten, 1 ¡2- response of a loudspeaker, it should be inch diameter holes. The I. Beranek. L.1... Acoustics, McGraw -Hill Book Co.. Inc.. N.Y.. 1954. page 129 equation (5.38) a careless design can very system satisfac- pointed out that loudspeaker/ enclosure 2. Ibid. page 123. equation (5.36). well spoil that response with spurious torily meets the requirements for an 3. /hid. up. cit. page 133, equation (5.45). resonances. economical, quality, high- fidelity sound 4. /bid., page 137 equation (5.54)

1 I I I I l f 1 9 e 7 Sound Pressure Responsi Q6 - - - - CLOSED BOX F5 OPEN BACK BOX cc DISTRIBUTED PORT 04 REFLEX ENCLOSURE I- tn 3

- - -- CLOSED BOX ...... OPEN BACK BOX DISTRIBUTED PORT REFLEX ENCLOSURE

Voce Coil Iwpedane

1 t 1 11

1 1 1 1 1 1 I I I I 1 1 1 1 1 t 1 1 1 I 1 1 1 1 50 100 600 20 100 1000 10000 20000 FREQUENCY (Hz) F REDUE NCY (Hz)

Fig. 6 - Sound pressure response and loudspeaker terminal impedance response for three types or Fig. 7 - Percent total harmonic distortion versus frequency in the loudspeaker enclosures. sound output of a small loudspeaker.

www.americanradiohistory.com 900 -MHz and 450 -MHz mobile radio performance in urban hilly terrain

F.A. Barton) G.A. Wagner Fig. 1 - Pittsburgh area map.

A field -test program to compare mobile radio performance in the 900 -MHz and 450 -MHz bands has been carried out in the Pittsburgh, Pennsylvania, urban and suburban area. Details of the equipment used, methods of measurement with preliminary results and conclusions are given.

IN THE NEAR FUTURE mobile com- Fig. 2 View of downtown area. - munication systems will be operating in the 806 -947 -MHz band, and a number of studies of such systems have been made. In cities such as Chicago, Washington, D.C., Philadelphia, and New York -the viability of practicable 900 -MHz systems has been well- established.1 4 Since the terrain in these studies was relatively flat, it is important to find out how such systems behave in more hilly areas.

The rolling hills, deep valleys and thick tree cover characteristic of Pittsburgh make it an appropriate region for additional study of 900 -MHz performance. Comparison with the existing 450 -MHz mobile band was desirable as a Fig. 3 - View of suburban area. general yardstick of performance. The basis of this comparison should also be IFig. 4 - Relief detaii of Pittsburgh area. made in terms of:

a) Voice modulation performance (qualitative only), b) Field strength variation (quantitative), and c) Data message transmission (quantitative).

To obtain a fair comparison of propagation between the two bands, it was necessary to develop 900 -MHz transmitters, receivers and antenna with performance equal to those already available in the 450 -MHz band. A brief

www.americanradiohistory.com description of this experimental equipment is included.

BASE STATION MOBILE TEST VAN

Topography of the Pittsburgh ERR 560 WATTS (57.5 awn) OR area 56 WATTS (47.5 <660 TRANSMITTER A 862 ARIA (db RECEIVER 100 WATTS GAIN ANTENNA Y A Fig. I is a map of the Pittsburgh area. The STATION (50) CABLE OR 10 WATTS (40 ,Ban) 862 MHz IDENTIFICATION/ LOSS average terrain elevation is 1000 feet VOICE/ DATA MULTI INPUT (312m) with peak -to -peak variations of TAPE DATA TRACK RECORDER 3 MIN DECODER,/ TAPE 500 feet (156m). The three rivers are TIMER PRINTER RECORDER bordered over much of their length by ERP 540 WATTS (57.3 5m0 steep 300- to 400 -foot cliffs (94 to 125m). TRANSMITTER e 8,25 56 OMNI Fig. 4 is a relief detail of a typical deep 454.55 MH: 0.55 GAIN ANTENNA RECENER 90 WATTS (49.5 &,n) L CARLE valley in the area. Because of this rough CT3 70BH 454.55 MHz LOSS terrain, there is essentially no line of sight CONTROL LINK EV 84 WATTS (49.2 &e) 4 ßy7 operation once outside the downtown TRANSMITTER C

SUB AUDIO L159.48 MHz area, RECEIVER C 3 and multipath propagation is the TONE 42 WATTS (46.2 mm) AIMICROlHONE1 177.48 MH: ANTENGAINNA rule tor the majority of the area. The DECODER CT 2007 concept of open radials and cross streets SUB AUDIO has no significance here, as terrain TONE ENCODER eliminates the possibility of any grid work of streets. Fig. 5 - Overall test system.

Test plan and equipment

Tests and measurements were ac- Frederick A. Barton, Leader, Advanced Development, Greg A. Wagner, Mobile Communications Product Mobile Communications Systems, Meadowlands. Pa.. Engineerirg, Meadowlands, Pa., graduated from Point complished with the use of a base station received an HNCEE degree from the Borough Park College, Pittsburgh, Pennsylvan',a, in 1970 with a transmitting site and a mobile receiving, Polytechnic London, England in 1944. He served as a BSET. He joined RCA in 1972 in the Product Design control, and instrumentation van (Figs. Radar Officer with the Royal Indian Navy from 1945 to Group and is a member of the engineering staff at 1947. From 1948 until 1955, he was a development Meadowlands. n January, 1973 he wastransferred to the 5, 6, and 9). Located on a large apartment engineer in the Communications Department of Redifon Advanced Development Engineering Group to design building on a 400 -foot (I25m) bluff Ltd., London. After two years with Central Rediffusion and construct the RF and control hardware used in this overlooking the city, the test base station Services as a CATV Systems Engineer in the field, he 900 MHz study. A 12 year background in VHF -UHF rejoined Redifon Ltd. as Section Leader in charge of HF amateur radio activity, including moon- bounce com- site was an availble site already in use by Transmitter Design. In 1960, he joined the Mobile munication provided useful experience for this program. several other radio systems. Communications Engineering Department of RCA and has had design and management responsibility for a variety of mobile products. A Senior Engineer in the Reprint RE- 20 -1 -6 Advanced Development Department, Mr. Barton became Final manuscript received March 21, 1974 leader of this group in April 1974. He is also Technical Th s paper was originally prepared for the Vehicular Transmitting and receiving equipment Publications Administrator for Mobile Communications Technology Professional Group 1973 Conference Systems, Meadowlands. Proceedings and is published by courtesy of IEEE. Both the 862- and 450 -MHz transmitters at the base station site were operated Authors Barton (left) and Wagner. unattended and were remotely controlled from the mobile van. A 150 -MHz control link equipped with subaudible tone equipment provided control of transmitter carrier on off and modulation on/ off functions. Audio for the test signal was on a broadcast type cartridge tape player fed to both base test transmitters.

A standard- product base station was used as the foundation for the 862 -MHz transmitter (Fig. 7) using the low -level, 150 -MHz multiplier output to drive a 450 -MHz tripler stage. The 150 milliwatt (21.8 dBm) output from the tripler drives a parametric transistor doubler to an output of 250 milliwatts (24 dBm) at 862 MHz. The doubler uses an inexpensive CATV transistor and is built with etched inductors on teflon board. Output from the doubler is fed to a three -stage thin - film hybrid amplifier module which

www.americanradiohistory.com dBm) for 20 -dB quieting.

The 450 -MHz receiver for the van was a standard -product unit with an optional rf preamplifier, producing a sensitivity of 0.25 µV ( -119 dBm) for 20 dB quieting. Mobile receiving antennas for both 450 and 862 MHz had 5 -dB gain.

Signal- strength data was recorded on a strip chart recorder for immediate Fig. 9 - Teat van. monitoring and on magnetic tape for more detailed analysis later (see Fig. 10). dBm) points.

At the end of a run, calibration was Signal strength measurement system rechecked. Maximum error even after several days operation was two dB. The The output level of the first stage of the stability of the system is due to the normal 455 -kHz amplifier in the receiver is linear voltage regulation of the receiver rf and with rf level inputs between 0.1 µV ( -127 i.f. stages, and the use of voltage regulated dBm) and 10 p.V ( -87 dBm). IC operation amplifiers with high feed- Fig. 6 - Base station used for test. back ratios in the remainder of the After linear amplification, the 455 -kHz system. produces the 10 watts (40 dBm) necessary signal is rectified in a full -wave detector. to drive the final -amplifier tube. This A logarithmic amplifier converts the dc multiplication and from the into a linear dB amplification output detector Chart records of the output from signal arrangement eliminates the need for a scale with a 40 -dB dynamic range. measurement system are particularly high -level varactor multiplier with its difficult to interpret in a raw form undesirable spurious products. The pen chart recorder can be connected because of: Operating at 100 watts (50 dBm) output, either directly to the logarithmic the tube -type cavity amplifier feeds an 8.5 amplifier output or to the playback head 1) Receiver internal noise at signal levels below dB omnidirectional vertical colinear of the corresponding fm record channel 0.25 MV ( -I19 dBm), and antenna, and including feedline losses on the tape recorder. The pen recorder 2) The rapid multipath variations typically was normally operated in the latter mode produces an effective radiated power of resulting in amplitude changes of 20 - 30 dB 560 watts (57.5 dBm). so as to provide continuous monitoring of at cyclic rates up to 100 times a second. the tape recorder performance. Initial calibration of the complete system was To provide a more readable chart provided by a signal generator connected recording, a 10 -Hz low -pass filter is Using inexpensive and commercially to the receiver input. normally switched in series with the chart available components, a standard - recorder input to dampen the pen move- product 450 -MHz receiver was modified In areas where signal strength exceeded ment. The chart therefore records signal for 862 -MHz operation for use in the IO µV ( -87 dBm) a switchable attenuator levels averaged over periods of ap- mobile van (Fig. 8). The addition of a hot - was connected in series with the antenna, proximately 100 milliseconds (Fig. 11). carrier diode mixer, bipolar rf stage and and the attenuation required to operate 862 -MHz rf selectivity produced a basic the receivers within the basic 40 -dB triple- conversion receiver with a sensitivi- dynamic range noted both on the pen Distance measurement system ty of 0.4 µV ( -116 dBm) for 20 -dB chart and the 450 -MHz voice channel of quieting. By using the local oscillator the tape recorder. Real time sampling of a signal level data multiplier output for two conversions, would not be suitable for statistical only one multiplier chain was needed. A The initial linearity of the logarithmic analysis as the speed of the truck must

single -stage rf preamplifier was added to scale was within about 1 dB for calibra - vary according to local traffic and road improve the sensitivity to 0.25 µV ( -119 tion at 0.1 µV ( -127 dBm) and 10 µV ( -87 conditions. This would cause a false

455 KHz 862 TO INPUT FROM 455 KHz FULL WAVE 862 MHz 862 MIl 423.75MHz RECEIVER IF AMPLIFIER DETECTOR PR EA MP SELEC- MIXER AMPLIFIER EXCITER TIVITY 6 431 MH 862 MHz 150 MHz RIFLER DOUBLER MULTIPLIER INPUT 438.25MHz 423.75 INSTRUMEN- L.O. MHz TATION LOGARITHMIC INJECTION SELEC- REGULATED TAPE AMPLIFIER 862 MHz 862 MHz TNITY INVERTER RECORDER OUTPUT 100 W (50 dbm) HYBRID POWER FILTER CAVITY AMPLIFIER SUPPLY AMPLIFIER MODULE *15 V MIXER, 23.75 TO 455KHz I.F 14.5 MHì 14.5 MHz STRIP CHART AND I.F. MIXER RECORDER AUDIO 110 Hz FILTER I-'-1.

measurement system. Fig. 7 - 862 -MHz transmitter. Fig. 8 - 862 -MHz receiver. Fig. 10 - Signal strength

www.americanradiohistory.com 50 FT. DISTANCE INTERVALS Table I - Tape Recorder Channel Assignments. the vehicle travels through the composite field the resultant amplitude at the receiver varies cyclically at a rate related Channel Type Information to the wavelength of the signal and the angle of motion of the vehicle relative to I Direct 450 MHr receiver audio output or the field. tape recorded notes

2 Direct 862 Mil, receiver audio output This phenomenon results in the well - M 3 fm 450 MHr signal strength ,rf i', ''77 4 fm 862 MHr signal strength known picket fence interference effect.

5 fm Distance marker signal Fig. 12 shows the signal strength varia- 454.55 Wit . t rv mm tion in both bands when the truck was CHART SPEED I .s, . -> driven at slow speeds alongside the RCA Fig. 11 - Typical received signal chart. weighting of the data, as the time spent in plant in Meadow Lands. The variation in high traffic or on poor roads would be both cases ranges from 20 to 30 dB in longer than that for other portions of the amplitude with the frequency of variation TO eV Nt1LTIPAIH SIGNAL VARIATIONS run. These areas may or may not be high - of the 862 -M Hz signal being ap- or low -signal areas, and would introduce proximately twice that at 454.55 MHz. 562 Mhh an unpredictable uncertainty into the I I'V. data. Real distance information, At 30 mi /h (48.km /h) the frequency of s however, is quite appropriate for use as a the fade rate is approximately 44 times/ tyV 1 sampling mode. A commercial distance at 454.55 MHz, and 88 times /s at 862 - 0 c measurement unit which produces output MHz. The deep fade duration is only a pulses at programmable intervals of small fraction of the fade cycle, and the driven distance was used to provide average signal strength is only slightly sampling triggers for analysis. Normally lower than the peak level. 454.55 Mtt

. I CV programmed to provide an output at 50- CHART SPEED 2i mm sec. foot ( I 5.6m) intervals, the distance information was recorded on one event Effects on voice channel of the chart recorder and on a communications Fig. 12 - Picket -fence signal variations. separate channel of the tape recorder. After an analog -to- digital conversion When the average signal strength was over a 50 -foot (15.6m) sample - and I µV ( -107 dBm) or greater, the picket IRT MD 711D AIM ate with hold intervals triggered from the fence interference effect was almost ' nrortT POeOE " ow sat en arT art distance channel of the tape recorder, the negligible. With an average signal of 0.25 varying dc outputs from the logarithmic tV ( -I 19 dBm), the interference was at amplifiers can be statistically analyzed. an annoying level but with little effect on T IrRVEIILL This analysis is still in progress. intelligibility. As the signal fell below 0.25 T I W Mt;nO :POP . OTYIT PVLM tV ( -119 dBm), there was a rapid Of WIT ONIT) o- 1 o o Antenna beam width degradation in intelligibility due to the L-MYOeeLAT110 DIe1T-J increased length of the bursts of receiver Calculations of the effects of antenna noise. Flg. 13 - Message format of test data. vertical beam width on signals received at various elevations encountered in the area of measurement show that these Effects on data reception should not be significant. This has been curries *sous P*OF1LE confirmed in the test measurements. One The data -signalling waveform consists of FLEVETION ELEVATION area, Saw Mill Run Boulevard, which is FEET YETENS a 1200 -Hz burst of tone. Data encoding is 500 ttaMNi.iiiiwi. tip¡N.I 405 only 0.6 mile (967m) from the base I...... NA,NMNII. achieved by phase shift modulation of the a r..N...... rNy N 510 station and is in the shadow of a 400 -foot r...... wli tone as shown in Fig. 13. IROo Nwlxí% 574 (I25m) cliff shows a relatively low but still 550 urn 5511 This is the rrrunrAtnr usable signal at 862 MHz. In both bands it was found that correct f..(wunrr'.uMNfy 'MI beam width Iloo .wf.ttlw111~u,1'V 545 combined result of vertical data messages were consistently received and shadow loss and is an extreme IDlO 551 if the average signal strength was I µV ¡I¡i¡¡..I example not likely to be encountered in 00o 751E ( -107 dBm or greater). Further detailed ¡ more level terrain. sw r '~/ > analysis of data system performance I!ÌÌÁ I/LVwy i111 results is underway.

Multipath propagation effects WO IIIM íi...... wÑ ass jI¡¡.;.m..:i 550 Owing to the limited elevation of the Signal strength prediction 150. r....°wi..í.wñ 05A ú 5 IIIII base -station antenna site and the uneven 700 20R s terrain, the received signal (in both BASE nRNn bands) in nearly all locations is the vector Meadow Lands Plant sum of a large number of reflected wavefronts of differing path lengths. As An elevation profile of the I9 -mile (30.4 Fig. 14 - Pittsburgh- Meadow Lands elevation profits.

www.americanradiohistory.com 4EARTM'S RADIUS PROFILE River. The predicted and measured Table III - Predicted and measured signals - base ELEVATION ELEVATION station to Harmarville. FEET METERS signals are shown in Table III. In both 905 locations reasonable agreement with the 1250',1250 prediction methods is Buffington Okamura demonstrated. Frey. Prediction Prediction Measurement 1200 -

Area signal coverage 454.55 MHz 4.5 gV 2.7 µV 5 µV ( -94 dBm) (-98.4 dBm) (-93 dBm)

862 MHz 3.2 aV 2.1 µV 3 NV ( -97 dBm) (-100.6 dBm) -97.5 dBm) Base -to- mobile

The two belt measurement runs shown in

Fig. 1 traversed the full gamut of terrain features that are characteristic of the Pittsburgh area; and, it has assumed that Mobile -to -base prediction a signal -quality analysis of these belts will provide a conservative estimate of the Predictions of 862 -MHz coverage to be signal variation pattern to be expected in expected on the inner belt (B2) using Fig. 15 - Pittsburgh -Harmarville elevation profile. the area enclosed within the belts. mobile transmitter powers of 10 watts (40 dBm) and 33 watts (45 dBm) are given in km) path from the base station to the The analysis method consisted of dividing Table V. The classification of signal Meadow Lands plant is shown in Fig. 14. the belts into four equal quadrants, and in quality was revised for reductions in turn, dividing the quadrants into ap- signal strength of 10 dB and 5 dB Table 11 compares Bullington5 and proximately 50 equal intervals. The respectively. Okumura6 predictions with measured results of this analysis are shown in Table median signal strength: IV. The minimum average signal strength Reducing the power level from 100 watts in each interval was classified as being in (50 dBm) to 33 watts (45 dBm) has one of three categories. changed coverage from 96% to 90%: Table II - Compares Bullington5 and Okumurae predictions with measured median signal strength. whereas a further reduction to 10 watts 1) Good: greater than the average signal level (40 dBm) results in a drop to only 63% of 1.0 tAV ( -107 dBm). Buffington Okamura coverage. For reasonable mobile -to -base hwy. Prediction Prediction Measurement I) Usable: between 0.25 µV ( -119 dBm) and coverage, a minimum mobile power of 30 the average signal level of 1.0 µV ( -107 dBm). watts (44.5 dBm) would be desirable for systems in this type of terrain. 454.55 MH7 1.6 pV 1.9 µV 3 µV 3) Not Usable: less than the average signal level (-103 dBm) (-101.5 dBm) (-97.5 dßm) of 0.25 p.V ( -I I9dBm).

862 MHz 1.1 µV 1.5 µV I µV (-106 dBm) (-103.5 dBm (-107 dBm) Tunnel propagation The S.E. quadrant has more steeply sided valleys than the rest of the area and has There are three major highway tunnels in resulted in worse coverage in both bands; Pittsburgh as shown in Table VI. the inner belt (B2) travelled over the It can be seen that a solid usable 862 - narrowest valleys of all (being worse than MHz signal was present throughout the Allegheny river at Harmarville the outer belt, B1). The shadowing effect length of these three quite differently is dependent on rate of change of eleva- located tunnel systems. The 450 MHz - Fig. 15 shows the elevation profile of the tion rather than the absolute changes signal was as much as 15 dB lower than 11 -mile path from the base station to involved; the latter changes are in fact the 862 M Hz-signal. Harmarville alongside the Allegheny fairly uniform over the whole area. By rapid changing of the 150 -MHz con- Table IV - Base -to- mobile measurements. trol signalling functions, loss of communication was shown within 500 -1000 OUTER BELT (BI) feet (156 - 312m) of entering and leaving PERCENT OF QUADRANT READINGS PERCENT OF BELT READINGS the Fort Pitt tunnels.

450 MHz 862 MHz 450 MHz 862 MHz 450 MHz 1 862 MNz 450 MHz 862 MHz 450 MHz MHz 842 The remarkable improvement in tunnel Good 73 54 83 52 Measurements 80 55 79 54 propagation at 862 M Hz is attributed to a Usable 21 30 15 42 not 18 32 18 35 more highly scattered signal illumination Not Usable 6 16 2 6 completed 2 13 3 11 at the tunnel entrance and multiple INNER BELT (B2) reflections along the tunnel walls.

PERCENT OF QUADRANT READINGS PERCENT OF BELT READINGS

450 MHz 862 MHz 450 MHz 862 MHz 450 MHz 1862 MHz 450 MHz 862 MHz 450 MHz 862 MHz Wind effects

Good 94 58 70 46 Measurements 100 85 85 62 days when parked Meadow Usable 6 42 28 43 not 0 15 14 34 On still at the Lands plant in an area with no other Not Usoble 0 0 2 10 completed 0 0 1 4 traffic movement, the received signal

www.americanradiohistory.com Table V - Mobile -to -base predictions. obstructions will closely follow those already experienced in 10 WATT (40 dbm) MOBILE AT 862 MHz - INNER BELT (B2) existing 450 -MHz systems with approximately 8-dB increased PERCENT OF QUADRANT READINGS PERCENT OF BELT propagation loss. READINGS 3)900 -MH7 multipath effects on both voice Good 29 8 Measurements 47 25 and data communiction will also be com-

Usable 29 42 not 38 38 parable to those in the 450 -MHz band.

Not Usable 42 50 completed 15 37 In addition, it appears that the 900 -MHz 33 WATT (45 dbm) MOBILE AT 862 MHz - INNER BELT (B2) band may offer some significant coverage PERCENT OF QUADRANT READINGS PERCENT OF BELT advantages over the existing band in city READINGS areas such as long tunnels and under Good 42 12 Measurements 0 34 elevated roads and railroads. Another Usable 48 70 not 42 56 marked improvement in coverage in the

Not Usable 10 18 completed 58 10 heavily noise -polluted downtown areas should result from the uniquely low -noise levels strength on both channels over a period proximately 0.15 µV ( -125.5 dBm) generated in this band. of several minutes showed little variation. signal. The antennas were then sub- stituted for the 50 -ohm load and degrada- Owing to the wide range of propagation losses However, on days with high gusty winds tion in quieting observed as various in terrain similar to Pittsburgh, the reuse in excess of 20 mi /h, variations of ±3 dB vehicles drove by. channel pattern in cellular systems at 862 MHz -and ±1 dB at 450 MHz will require extreme care in design to avoid interference were observed at roughly half a second Significant quieting degradation at 450 intolerable between cells sharing common channels. cyclic rate. Base station and /or mobile MHz was observed with some vehicles antenna motion under these wind con- (particularly certain motor -cycles). This Looking ahead to the hardware ditions does not appear to be great degradation typically reduced quieting to requirements for the new enough to account for the magnitude of 8 -dB. At 862 MHz, although no band, recent component tehcnology developments the signal variation, and it is believed that measureable effects were observed on a have made practicable the design the phenomenon may be due to patterns Ballantine meter, some mild audible of mobile and base stations comparable in of foliage movement with wind gusts. clicks were observed. performance to existing 450 -MHz equip- The Test Van which had its motor run- ment. Interference ning at all times had no measurable or audible effect. Observation of the noise Acknowledgments During the course of hundreds of miles of spectrum appearing on a spectrum measurements, degradation of audio out- analyzer connected in turn to both 450 - The authors acknowledge: put in weak signal areas due to ignition MHz and 862-M Hz antennas confirmed inteference from either our own or other that vehicle interference frequency com- I ) Antenna Specialists,Company Engineering Department, vehicles could not be positively identified ponents could be identified up to ap- 2)Communications Research Group of RCA on either the 459.55 -MHz or 862 -MHz proximately 600 MHz but not above. Laboratories, channel. This is due to the fact when that 3) Power Products Engineering, RCA Elec- in motion the multipath fade effects tronic Components, and closely resemble impulse noise. To Conclusions obtain 4) RadioFrequency Power Engineering, RCA a better quantitative picture the test van Solid State Division. was in a busy downtown street for an Our preliminary measurements have con- hour. firmed findings of previous investigators In addition, R. Ferrie and J. Kaltenbach that: of Mobile Communications Design Fite rf inputs to the receivers were initially Engineering, Meadow Lands, Pa., terminated in 50 ohms with signal 1) 900 M Hz city -wide mobile dispatch systems provided considerable support in this will be generator inputs provided practicable in the near future, not program. by a high only in level areas but also in more un- attenuation T coupler. The signal dulating terrain such as Pittsburgh. generators were adjusted to provide a 10- 2) The propagation effects of changes in eleva- References dB quieting input corresponding to ap- tion due to either natural or man -made

. Black, D.M. and Reudink, D.O., "Some characteristics of mobile Table VI - Tunnel propagation. radio propagation at 836 MHz in the Philadelphia area, "IEEE Transactions on Vehicular Technology - May, 1972. ."Examination of the feasibility of conventional land mobile operation at 950 MH7 ", FCC Research Division Report R- Distance Minimum Minimum 7102. From Base Length 450 MHz 862 MHz I.ynk. Charles N., "Multi -user dispatch system "IEEE Tunnel Station (Yds.) Signal Signal Vehicular Technology 23rd Annual Conference - December. 1972.

. Shefer, J., RCA Labs, Princeton, New Jersey, "Private Fort Pitt 400 ft. (125m) 1200 0.5 µV 1.0 µV Correspondence." May 2, 1973. underneath (1097m) 13dBm) (-I (-107dBm) . Bullington, K.. "Radio propagation at frequencies above 30 Mc", Proceedings Liberty 1.4 miles 2000 0.8 yV 0.5 µV of the IRE, 35 (October, 1947) (2.25 km) (1829m) (-109d Bm) (-I 13dBm) .Okumura. Y., Ohmori. E., Kawano, T., and Fukuda, K., "Field Strength and Its Variability in VHF and UHF Land- Squirrel 5.5 miles 1370 0.3 uV 1.5 uV Mobile Radio Service. Review of the Electrical Com- Hill (8.85 km) (1253m) (-I17.5dBm) (I-103.5dBm) munication Lahormory Vol. 16. pp. 825-873, Sept: Oct., 1968.

www.americanradiohistory.com Computer program architectural design

PREEMPT1ON PRIORITY WITHIN TASKS for weapon system radar control LEVEL PREEMPT ION LEVEL I F

2 1

T. Mehling 3

in the of large - This paper presents an example of the methods used developing structure 16

scale Command and Control programs. First a description is presented of the basic 1 elements or building blocks used in such programs; then the methods used in developing 2 the AEGIS AN /SPY -1 radar control program architecture are described in terms of these 3 building blocks, with examples of the factors influencing the architecture. Finally, the tools 2 `S H with examples of their use. The of the computer program architect are discussed 6 processes described will vary somewhat as a function of the tasks to be performed by the program, but will generally be similar to those described in this paper. 16

2 e

3 4 and programs performing complex functions FOR LARGE command control 3 s RS computer programs, the development within stringent timing requirements. An 6 process has only begun when the example of this complexity is the functions to be performed have been AN /SPY -1 radar control program 16 identified and defined. The next critical package for the AEGIS Weapon System. 1 2 is the The architectural development process step the translation of functional 3 design into the computer program system for such a complex system can best be structure that is, the computer understood by a study of the basic 4 `s program architecture. Thorough building blocks. 6 C software architectural development is es- 16 pecially important for multi- computer For large, complicated computer programs, a modular program design Fig. t - Executive priority structure. Thomas H. Mehling, Signal and Data Processing approach is used with three basic classes Moorestown, NJ, received the BEE Engineering. MSRD, of elements: 1) the task -performing sub- and MEE from the University of Louisville in 1961 and Task -performing module 1962 and has taken additional graduate work in systems programs (or modules), 2) the executive. engineering at the University of Pennsylvania. On joining and 3) the data base. In the simplest The first of the three building blocks in in 1962, he was a member of the BMEWS Systems RCA terms, the executive controls the flow of Engineering Department, developing requirements for the modular program design are the task automatic operability testing of the BMEWS radars. tasks between the program modules, subprograms (modules). In general, a Since then his responsibilities have been in the specifica- which in turn communicate with each module performs several functionally tion of requirements for real time computer programs, as well as their implementation and testing. Since the late other and the outside world via the data related tasks. The module itself may be 1960's he has been assigned to the AEGIS Program; he base. further subdivided into procedures for the participated in the development of specifications (subroutines) to perform specific tasks at AN /SPY -1 programs within AEGIS at the AEGIS Land Based Test Site; he is currently responsible for AN /SPY- A modular program approach permits one or more points within the module. In 1 control program integration and testing aboard the partitioning of the job into manageable addition, the module may be entered at Sound. received Chief Engineer's USS Norton He work packages for implementation of the Technical Excellenun Awards for his work on the SAM -D guidance system flight tests. and for his work on program and ease of program Table I Module entries. integration and test of AN /SPY -1 control programs in maintenance. Thus, the performance re- - 1974, respectively. 1967 and quirements for the individual program components must be developed along /iure Purpose with the interfacing definitions, a process common to development of complex functions. As hardware implememented Initialization Initialize module data and with complex hardware, the mere input output devices. satisfaction of program module perform- Error Process module -related errors. Successor Process a request from ance requirements does not guarantee the another module. performance of the entire system unless Message Process a request from another module with input data these requirements are carefully con- in temporary storage. sidered in the context of interface re- Periodic Perform periodic functions. quirements and total system perform- Huffer complete Processing due to termination of Channel complete input 'output ance. functions related to the module.

Reprint RE- 20-1 -14 Final manuscript received May 2, 1974.

www.americanradiohistory.com more than one point to perform a specific ing erroneous outputs to the interrupted tasks and any tasks of the same priority task. This is accomplished either through module. requested earlier. procedures independent of those for other entries, or by sharing some or all In AEGIS, both classes of time criticality procedures from other entries. In Executive program are present. To handle such re- AEGIS, for example, each module has quirements, a priority system has been seven classes of entries (Table I) which are In general, the executive allocates com- implemented which allows preemption as called via the AEGIS Tactical Executive puter processor time, memory, and in- needed and provides a priority structure Program (ATEP). put /output facilities for use by the task - for non- preemptive tasks. The executive performing modules. The degree to which structure contains two levels of task All modules have access to common this is done depends largely on the priority one for the preemption level, subroutines as well as to their individual computer design. In AEGIS, all com- and another for its execution within that procedures. The common subroutine is a puters used are Univac AN/ UYK -7's preemption level. Fig. 1 showsa represen- special class of subroutine usable by more which have features that tend to dictate tative Priority structure where four than one task -performing module; requirements for the executive program. preemption level. Fig. 1 shows a typically, it is used to compute trigono- This computer has a set of privileged representative priority structure where metric functions, logarithms, and other instructions that may be executed only four preemption levels exist with sixteen mathematical functions. Since these sub- when it is in its executive state. Among of priority represent the higher priority routines are common to several modules, these are instructions which use memory values. The tasks listed at right are shown care must be taken to prevent inadvertent protection facilities to exercise direct at the preemption level and priority interference between the user modules control over memory access by task within that level at which they are to be through these subroutines. If one task mod u le. executed. Fig. 2 is an example of the module is allowed to interrupt operation operation of this priority scheme. of another task module using a common In this process, the executive must subroutine, errors are likely in the allocate processor time to task modules The top line of this time -line diagram resulting output from the subroutine to for timely execution of the tasks required. shows the occurrence of events which re- the interrupted module. If the subroutine This implies a means of scheduling and sult in request for use of the computer's uses the same memory to store in- executing task module entries based on central processor unit (CPU). The next termediate results within its com- task priority and individual task reaction line shows the content of the priority putations, these intermediate results will time. Reaction time is categorized in two queues as a function of time for tasks be overwritten if the subroutine is in- classes of requirements, preemptive and awaiting CPU time. The third line shows terrupted and reentered prior to finishing non -preemptive. The most critical time usage of the CPU at any given time. The the original computation. Such sub- requirements demand that tasks in last line shows which tasks are suspended routines are called non -reentrant in that process be temporarily suspended until at any time, awaiting resumption of CPU they may not be successfully reentered the short- reaction-time task is per- availability. At t =0, task A is being during the performance of computations. formed. At the termination of this time - performed by the CPU with tasks B,D, critical task, the suspended task is picked and C awaiting service. Prior to comple- The subroutine may be successfully used up at the point of suspension. This tion of task A, a request for task E occurs by modules which interrupt each other preemption process is referred to at a higher preemption level, causing only if care is taken either to store all throughout the remainder of the discus- suspension of task A. Shortly after this intermediate data in memory individually sion. event, task F occurs, with the same assigned for a particular subroutine call, preemptive priority as E but with higher or to keep this data in registers which the The second general class of time criticali- priority within that preemptive level. By executive saves and restores during these ty is referred to as nonoreemptive, in the rules, the preemptive priority is not interruptions. Such subroutines are which tasks in nrogress are completed higher than that of the task being ex- referred to as reentrant subroutines prior to execution of a new task. 1 he ecuted; therefore task F is placed in the because of their ability to be interrupted execution of the new task, then, will occur priority scheduling queue, in this case at and subsequently reentered without caus- upon completion of all higher priority the top of the queue. When task E is

TIME complete, task F is begun and deleted 1., from the priority queue since it is also at a

REQUESTS E F G H I FOR higher preemptive priority than the P(Ii2) P(1,11 P13.51 P(2,9.1.(I,2) TASKS suspended task A. When task G is re- it is PRIORITY 6 9 " F quested, placed third in the priority P(3,2) P(32) PILI P13,21 P(3,21' P13,5)-P13,9 QUEUE AT ANY queue since it is not preemptive. Also, P(3,5) P(3)Pg21 P13?) P(3,5) P(3,3) P(0.6) GIVEN TIME (HIGHEST G C since it occurred after the request for D P(4.6 P(4,6) P(D3.5) P(4,6) PRIORITY %3,51 P(4,6) AT TOP) and is of the same overall priority, it is P(.%1 P(ó,6) scheduled after D in first -in first -out CPU 4v % í//' / r. %/ ! %// % ! y : ' % % ! % % % UTILIZATION // / /' / 0 /©/%/i%//// // v/i%/i /// % // ///A (FIFO) fashion. At completion of task F, .ii ..,. task A is resumed since the queue con- SUSPENDED iia©ii.. TASKS / tains no other requests _ of higher ;5*///// ,0iiiiiiiii. preemptive priority.

P(X,Y) .PRIORITY WITHIN PREEMPTIVE PRIORITY X When task A is finally completed, tasks B Fig. 2 - Examples of AEGIS executive priority structure operation.

www.americanradiohistory.com RAW GYRO GA SPECIAL YRO BEAM $REQUESTS operating constraints, and system data this form of data includes Operational ALL SEARCH SEARCH RADAR BEAM BEAM BEAM OUTPUT such as target tracking data, coordinate Readiness Test System (ORTS) data MANAGEMENT TYPE SCHEDULING REQUESTS STABILIZATION REQUESTS FORMATTING REQUESTS conversion matrices, and display data. which includes test data and evaluation TRACK BEAM REQUESTS as well as display messages for COMMANDS criteria TRACK In the context of AEGIS software, there maintenance action. Since ORTS and MANAGEMENT RADAR -1 the same two HARDWARE are three general classifications of core - SPY software share TRACK resident data: common data, private computers, the use of the non -core- REQUESTS !RADAR RETURNS data, and temporary data. Common data resident data base is a factor in the TRACK STABILIZED w PROCESSING TRACK DATA UNSTABILIZED RADAR consists of that 'data shared by two or architecture of the AN /SPY -1 software RETURN TRACK DATA PROCESSING NEW TARGETS more task -performing modules. General- even though the normal SPY -I functions

DETECTION DETECTION DATA ly this implies that any of these modules do not require this form of data base. PROCESSING sharing the data may both read and modify the data; although in many cases, Fig. 3 - Primary radar control loop functions. modification of some portions of the Development of the AN /SPY -1 common data base is restricted by design program architecture to only one of the modules. Private data, and D are given use of the CPU. Prior to in contrast, is only accessible by the At this point, the development of the completion of task D, task H is requested. module with which it is related. It is used architecture for the AN /SPY -1 radar This again causes preemption of task D to maintain constraints and working data control program will be discussed. This because of the higher preemptive priority needed within the module, whereas com- radar system, developed for the Navy as of H. This time, however, a new higher mon data is used for passing data between part of the AEGIS Weapon System, is a preemptive priority, task I, occurs prior modules. Temporary data (or storage) is multi -function array radar capable of to completion of H, and H is also generally used to pass data from moaule maintaining track on a large number of preempted. When task 1 is complete, task to module as a message without requiring simultaneous targets as well as sur- H then task D are completed, and task G a permanent allocation of memory to veillance of the radar volume for begun. hold the specific data passed. This is done appearance of new targets. In this system, by dynamic allocation of storage the radar receives commands from the In some computers, a multi -level locations by the executive program upon radar control software, directing the hardware interrupt system is used with request of a module. This module in turn energy relative to the array separate registers for each level of task to transmitted places data in the allocated memory area antenna and the required waveform, and be performed. This can greatly simplify and then, via the executive, passes the processes the returning signal to the role of the executive, sometimes to the then data to another module, requesting it to determine the instantaneous measured point that the executive function is do the required tasks as determined by of a target or targets within the almost entirely assumed by the hardware position the contents of the temporary data passed beam in the presence of noise, in- interrupt system. In this case, the ex- to it. terference, or clutter. The programs in the ecutive program is effectively im- radar control computers generate corn - plemented in the computer hardware. In In this case, the data is accessible to the mands to the radar equipment within the AEGIS, on the other hand, the executive sending module only up to the point of constraints of the capabilities of that function is nearly all implemented in sending the message, and by the receiving equipment and user requirements. These computer software. The AN /UYK -7 module only after it has been requested to programs also initiate and maintain computer has four interrupt levels, but process the request. Upon completion of tracks on targets reported by the radar these are used strictly to perform ex- this processing, the temporary storage is ana provide the users with ecutive functions and are not available equipment, generally returned to the pool of tem- requested track and detection data and for use by the task -performing modules. porary storage locations and is not other display information. The primary The interrupt levels are I) hardware - accessible to any module until the process functions of the radar control software detected hardware faults, 2) hardware - is repeated. Temporary storage can also listed below, with a brief description detected software errors and periodic are be used as a temporary work area within a of each. (See Fig. 3). clock interrupts, 3) input /output -related module which is released upon exit from interrupts, and 4) program- requested in- the module. This is generally not done terrupts by which the task modules re- because of the overhead times that the quest service of the executive program. I) Radar scheduling the process by which executive would take in allocation and is allocated on a basis Although this represents a priority in- radar time priority deallocation of the temporary storage; within the constraints imposed by the terrupt structure, the priority is related instead private data is generally used. radar equipment, such as transmitter strictly to those priorities of importance average power and phase shifter switching - to the execution of the executive; it is in rate constraints. This process must also Another whole class of data remains and no way directly usable for task -module meet system reaction times required to this is the non- core -resident data base. initiate and maintain track for the various priority scheduling. The non -core -resident data base normal- classes of track. ly is required whenever the overall data- 2) Search management - the process by Data base base size is too large to contain in on -line which search and special beam requests to the radar scheduler are generated. The core memory and is also amenable to generation of search requests includes the The data base consists of all data with lower -speed access usually available from search beam pattern generation within the which the programs must operate. This the off -line storage devices such as disk, constraints of the requested search data includes doctrine and rules defining drums, and magnetic tape. In AEGIS, coverage defined by the radar operator.

www.americanradiohistory.com The special beam requests, such as acquisi- 14) Operational Readiness Test System redundant data needed, minimize the tion and ORTS special test beams, will (ORTS) - though not specifically part of amount of intercomputer input/ output result in generation of a request(s) to the the radar program, control ORTS traffic and delays, and maintain a scheduler if the requests are allowed. monitors operation of the radar equipment. Since it shares the computers used by reasonable balance of computer load and 3) Track management - the process by which the radar control programs, the needs of memory utilization between them.. At this request for track dwells are made to the the ORTS programs must be considered in scheduler, including determination of point, it should be stressed that there the architecture of the AN -I priority of individual /SPY control should be reasonable memory and timing dwells and the se- program. quence of the requests. estimates for the individual functions 15) Background self test - the process per- 4) Output formatting the process by which (and in many cases subfunctions) before formed in spare computer time, used to this allocation process can reasonably be dwell requests from the scheduling func- verify proper operation of the computers. tion are converted to commands to the made. In many cases this estimation radar equipment. process may require detailed algorithm An examination of this set of functions 5) Radar return processing -- the process by develoment of particularly critical tasks shows that the first eight are the primary which return data from the radar equip- (i.e., those that could require large functions in the radar control loop and ment is pre -processed to determine amounts of computer time or memory) to presence of valid data resulting track and impose the most severe time requirements guarantee the feasibility of performing search data are then forwarded to the track and timing restrictions on the total and detection processing the tasks within the available computer functions. program; obviously these functions will resources. 6) Beam stabilization -the process required tend to dictate the design. Fig. 3 shows the to account tor the heading, pitch, and roll interrelationships between these motion of the ship in generating - A good example of such a task is the beam functions and related equipment, in- steering commands and computing cross -gating process, a part of the detec- cluding the radar equipment and gyro. stabilized tracking -error data derived from tion processing function. This process radar returns. requires correlation of a new detection A preliminary analysis of the search 7) Track processing the process by which report with existing track data to verify requirements and track capacities for the track- return data is used to update target that the detection is not an existing track. tracking data (smoothes position and AEGIS Engineering Development Model In the brute -force approach of correlating velocity), track look -rate, and waveform (EDM -I) indicated that two AN /UYK -7 determination. The process also includes the detection with every track in the computers would be required to perform initiation of new target -track data based on system, the required processing time all required tasks for both ORTS and the output of the detection processing would tax the capacity of several com- function. radar control programs. This version of puters, assuming normal search under a the radar uses one array antenna covering 8) Detection processing - the process by full load. The process can be greatly one quadrant of space about the radar, which firm detections derived from search simplified by segmenting the volume dwells are correlated with existing tracks and is used to demonstrate the radar about the radar into bins of given azimuth using a process called crossgating to performance. The AN/ SPY-I radar to be and elevation, and then maintaining a determine if the detection is a valid new used aboard ship for anti aircraft defense detection or a redetection of an existing range- ordered list of all targets within will make use of four array antennas track. 1f it is a valid new detection, track each of these bins. In this way, only processing is requested to initiate track on covering the full volume about the ship targets in the bins in the near vicinity of the new target. If correlation with an and will require additional computers. the detection need be examined for possi- existing track occurs, the new detection is All further discussions of the architec- dropped from further consideration. ble correlation. tural development presented will be based 9) Track association the process by which on the Engineering Development Model redundant tracks, which occur as results of By use of the optimum bin size for the which requires two computers. It was merged tracks and crossing tracks, are maximum target motion expected and for found that the radar- control loop itself removed from the track data files. This isa ease of computation, the time required low priority frunction accomplished at a could not be implemented solely within for this process was reduced ap- low rate. one computer because these function proximately 30 -fold to a quite acceptable I0) Switch aciton/ display processing the require more processing time than a - load of computer usage even when process by which radar- control operator single computer can provide. requests are processed and displays are considering the additional processing re- generated for the operator, including quired for maintaining the lists of targets various symbolic displays, raw video dis- in each bin. In this case, the additional Two -computer configuration plays, and alphanumeric readouts. memory required for both the program 1 I) User service - the process by which radar Development of the software architecture and related cross -gate bin tables was very data (including raw radar track data) is within the two- computer configuration acceptable, making this an effective sent to users such as the command and time/ memory tradeoff. control weapons segments. This function requires allocation of tasks between the also includes reporting results of all other computers and then to the individual requests by these segments and general task- performing modules. The allocation status reporting. criteria include the ability to perform the Program efficiency 12) Recording - the function by which data is required tasks within timing and capacity extracted and recorded on magnetic tape constraints, plus several other factors As in most modular programs, program for post analysis of radar performance (primarily for purposes of testing). which determine the efficiency of the efficiency is enhanced if as much process- design and possible growth capability. In ing is done within a module as practical 13) Load evaluation the process which maintains radar and computer load such a two -computer configuration it is prior to giving control back to parameter information for the system desirable to partition tasks between the the executive. This fact leads to two operators. computers to minimize the amount of different actions. The first of these is to

www.americanradiohistory.com SPECIAL GYRO BEAM RAW REQUESTS data for other radar data users, but not GYRO DATA ALL for the actual communication of this data SEARCH SEARCH RADAR BEAM BEAM BEAM OUTPUT MANAGEMENT TYPE SCHEDULING REQUESTS STABILIZATION REQUESTS FORMATTING is REQUESTS (which accomplished by other !TRACK BEAM modules). The search- management 11 REQUESTS COMMANDS module generates search beam requests TRACK MANAGEMENT RADAR and other special beam requests whenever HARDWARE the is I TRACK scheduler capable of processing REQUESTS IRADAR The modules make the RETURNS them. four that up TRACK STABILIZED UNSTABILIZED critical part of this radar control loop are PROCESSING TRACK DATA TRACK DATA RADAR RETURN radar scheduler, output formatting, radar PROCESSING NEW TARGETS return processing, and track processing. DETECTION DETECTION DATA PROCESSING The other two, gyro data update and search management, are of lower impor- COMPUTER I COMPUTER 2 tance on the short -term basis although Fig. 4 - Primary radar control loop functions by both are necessary for the overall opera- computer. tion of the radar.

combine functions assigned to the commands to the radar equipment (using The four critical modules are designed to on modules; the second is to lower the rate at gyro data provided by the gyro -data- operate an amount of data equivalent which modules are entered by having the update module which periodically to an amount of radar time called a module process data in batches on each samples the ship's heading, pitch, and scheduling interval which has a nominal value Ts. Every Ts milliseconds, the call rather than a single data item. This roll). When formatting these commands, scheduler is called by the executive to batch processing of data is particularly the output- formatting module passes suf- generate a variable number radar beam advantageous in the radar loop where ficient data to the radar- return- of for a 2 Ts to 3 Ts in hundreds of beams must be generated and processing module to identify the purpose requests time span the The is processed each second; however, there of each beam, plus control data to enable future. number of beams limited by the number of dwells the radar is capable are drawbacks to this approach. The the radar- return -processing module to within its hardware prime concern with the radar control loop control both the radar input and output of executing constraints and by the amount of time is the ability to close the radar tracking channels. The radar -return -processing required by the four modules to process loop in the software for the maximum module performs its assigned function as number of beams without tracking rate required. The next major well as beam stabilization for track return that software overloads on the the short -term basis. step, then, is to determine the size of the data. It uses matrices generated by output Subsequently, output formatting will for- data batch to be processed in these formatting in the generation of corn - mat that number of requests a batch. control loop modules, and the makeup of mands for these track beams. as Only when the return from the last beam these modules. is in memory will Track and detection data are forwarded of this group available the radar- return -processing module be Fig. 5 shows the modules that comprise to the track -processing module, and those the radar control loop. The radar returns from special ORTS dwells are called by the executive to process Track processing will, in turn, scheduler module basically performs the passed on to the appropriate ORTS returns. radar scheduling function, passing all module. The track- processing module process all return data derived from that interval. The restriction on the number of required request data to the output - performs the track processing, detection in module formatting module. This module per- processing, and track management beams any interval based on forms the output formatting function and functions, passing priority- oriented re- run time should be such that the combined execution time of the critical loop that portion of the beam stabilization quests for track beams to the scheduler modules in each computer is less than Ts; function required to generate steering module via data tables. It also extracts otherwise short -term overload (if not long -term overload) would be possible.

SPECIAL BEAM REQUESTS Other modules in the radar control OUTPUT SEARCH SEARCH RA BEAM portions of the program have been MANAGEMENT REQESTM"S SCHEDULER REQUESTS FORMATTING restricted in execution time so that they

TRACK RADAR need not be preempted by critical loop REQUESTS HARDWARE modules. This restriction implies that the

RADAR amount of processing per call is limited to TRACK AND TRACK RETURN PROCESSING DETECTION DATA PROCESSING this maximum value. If necessary, the

GYRO module may call itself to complete the rest DATA GYRO UPDATE of the required task after the higher DATA priority modules are executed. Although will give GYRO preemption quicker response COMPUTER I COMPUTER 2 times to the critical loop modules, there Fig. 5 - Primary control loop modules. would be a heavy penalty in executive

www.americanradiohistory.com overhead for the use of the preemption satisfied by selecting a scheduling interval Even so, if the entire process for all tracks feature, as well as possible complexities of 100 ms. Though not specifically stated, is performed in one pass, the execution caused by data being changed by one the priority of these modules within the time could be in excess of that allowed for module that preempts another. Based on same preemptive priority level as the modules not in the critical loop and not these constraints, Fig. 6 shows the timing other radar control modules is higher preemptable. This module is called relationships that could exist in both than that for any of the remaining periodically at a very low rate to perform computers for the execution of these modules with the possible exception of this function. If it cannot be completed in critical loop modules. the user -service modules for the com- the allotted time, processing is broken olf mand and control interface and weapons prior to exceeding this time, and the This atasram represents a time history of interface (which both have the same module calls itself at the same lower the events occurring in each of the com- priority as the critical loop modules). priority to continue processing where it puters and in the radar equipment. For This is required to ensure the required left off, thereby allowing the higher example, at time = to a periodic request is reaction times for new high priority priority critical loop modules to be ex- made to the scheduler in computer I , but tracks and to prevent problems with a ecuted without undue delay. since another module is being executed at staleness of data when using it as part of the present time, the radar -scheduler the launch/ intercept requirements of the module is delayed until the area labeled weapons segment. Since the processing Switch -action /display processing module RS /A (radar scheduler/ interval A) is time required by these modules is very At short relative to the critical loop modules, shown. the end of the required In the switch -action /display- processing in the a request is the example of Fig. 6 is still valid. processing scheduler. module, the most time- consuming re- sent to computer 2 where an I;'O delay is quirement is the updating of the target encountered. In computer 2, there is a In computer I (where the scheduler and symbology for all tracks at a minimum further delay - -- again to await execution track- processing modules reside along rate of once every two seconds. By use of a of another module labeled OF/ A (output with the shared track file and related 16 /s periodic entry to this module to formatting interval A). Output for- tables) the track association function, update approximately 1/ 32 "" of the track matting generates commands to the radar switch action/ display processing func- symbology and any other indicators as for the interval labeled A on the bottom tion, and user service function are then to they require it, the peak execution time is line starting at to + 2 Ts. At the end of this be implemented as well if time and held to an acceptable value. Entries interval (t + 3 Ts) is computer 2, the memory permit. Since there is adequate resulting from operator actions are in- radar -return processing module will time and memory to do this, these frequent and require processing time in process the return data from interval A functions are implemented there because amounts well within these limits as well. (labeled RRí A). of the availability of the track tile and related data in the common data portion When processing is complete, it will send of the data base in that computer. The Other modules the return data to C I where the track - modules are called track association, processing module will process the return switch action/ display processing, com- Of the remaining radar control -related data again after a delay to complete mand and control user services, and functions, recording modules exist in another module's execution. As shown, weapon control user services. both computers, as well as background track processing for interval A is always self-test modules performing the completed prior to execution of the radar functions of the same names for both scheduler for interval E. It is then possi- Track association module computers. Physical extraction of the ble, for example, for a track scheduled in data is performed by a common service interval A to be rescheduled in interval E The track- association module makes use routine which does the required extrac- after updating the track and a delay of of the cross -gate bin lists mentioned tion of data upon request of the in- four intervals (4Ts). By allowing another earlier to speed up the correlation process dividual user modules processing the interval Ts for possible scheduling con- required to eliminate redundant tracks required data. The recording modules are flicts, it can be seen that a loop closure which occur when track targets merge. physically responsible for control of the time of 5 Ts is easily to be expected, provided the priority rules for the required high data rate track support this short scheduling delay. Although the CI scheduling priorities are not discussed RS/A I 11-13B T17/71' RS/C I RS/D TP/A KS/E a3Ts here, this condition is also met. Therefore Ob ¡ Ts 2Ts I/t7 DELAY I/O DELAY if a loop closure time of Tis required, the scheduling interval (Ts) should ideally be C2 OF/ A RR/ Y OF/ B RR/ i* OF/C RR/A

1 T/ 5 or less. For sake of efficienty the T/ 5 co value is approximately the value used for LAST BEAM Ts. COMPLETE

XI Y A C RADCR r If, as a hypothetical example, a 2/ second EQUIPMENT closure rate is the maximum required, the loop closure time of 500 ms would then be Fig. S - Critical loop timing.

www.americanradiohistory.com actual recording of the data on magnetic to execution of these subroutines and poses of determining queueing delays and tape. The load -evaluation module resides again preemptable upon their comple- computer load; the need here is for the in computer l and periodically computes tion. arrival rate of targets and the computer load status (radar load, computer load, processing times to perform the detailed number of tracks, etc.) based on data algorithms. provided by other modules. With the final set of modules briefly outlined, the requirements for both the It is the simulation of queueing delays radar control program and the ORTS and reaction time that will prove to be programs are met and a Operational readiness test system reasonably good most useful, especially when the amount balance of memory and timing is of computer processing required is high The only remaining function to be im- achieved in both computers. This discus- relative to computer capacity, or where plemented is ORTS. This, in fact, is not sion is only representative of the total peak -load conditions exist which would just a single function but includes many architectural task; many details have create possible overload situations. Using test -related functions required to monitor been omitted. For example, as mentioned this simulation, the priority structure of in the the operation of the radar and notify the executive discussion, most modules the program can be evaluated under have maintenance operator of failures. By several entry points although those expected operating conditions. In some nature of the function, it is difficult to mentioned in the architectural develop- cases where the detailed algorithms of the partition some of these tasks to fit the ment are the most important ones. simulated modules affect the amount of execution time limits for non - processing performed by other modules, preemptable modules. Also, these these algorithms must be included in the modules require a relatively large amount Tools of the software architect simulation. The radar scheduler module of memory for data and program, but on is such a case since it determines the type the average require a small amount of So far in this discussion the basic tools and number of radar beams to be computer time (although sometimes in used have been computer load and generated. In developing timing es- relatively long periods). Because of the memory estimates, input/ output traffic timates, those functions which are most core requirement, all ORTS functions are analysis, and elementary loop -closure frequently used and of higher priority implemented in computer 2 which has timing verification. As in any other should be done most carefully. As the relatively light memory usage from the systems engineering task, the develop- development process continues, the es- other modules there. In those ORTS ment of a viable software architecture is timates should be updated to reflect latest modules where execution times could usually the result of an iterative process in timing values as detailed algorithms are cause problems with the critical radar which the interactions resulting from derived. If necessary, the priority struc- control modules, the priority is such that proposed changes in the design can be ture should be modified to meet reaction these modules may be preempted by any evaluated. One of the most useful tools time requirements and queueing other module except possibly other available to the software architect is problems. Excessive peak loads may well ORTS modules or the background self - simulation. The forms of simulation point out the need to develop algorithms test module (which by its nature must be available can be used to evaluate the that will permit the modules to limit the the lowest priority module in the com- detailed algorithms used within the peaks, thereby preventing failure of the puter). program, as well as queueing delays and program and the overall system of which the general operation of the priority it is a part. structure for the modules. Using either Those ORTS modules that interface with type of simulation, a model of the the radar control modules are for the program and its environment must be Conclusion most part non -preemptable, but of lower developed. If this model is not developed priority than the non -preemptable radar carefully, the end result of the simulation The software architecture developed us- control modules. This avoids the problem may be nearly useless. For this reason, ing the techniques and tools described in of data being changed by the higher great care should be taken to include all this paper has been successfully priority module which would otherwise relevant characteristics in the simulation demonstrated for the AN /SPY -1 radar preempt the ORTS module, creating model without including unnecessary control programs in a series of rigorous unnecessary problems if preemption is details which greatly affect the amount of tests. This testing has included extensive not really required. Those modules that work required but do not affect the land -based AN /SPY -1 radar perfor- are preemptable must be designed to results. mance testing as well as AEGIS system - prevent this type of problem. If common level tests, including the Weapons Con- subroutines were used by both non - trol Systems and Command and Control. preemptable and preemptable modules, In the AN /SPY -1 radar program The final days of this testing included a it would be possible for a subroutine development, a good example is the use 48 -hour "fast cruise" during which con- being used by the preemptable module to of target motion parameters in simula- tinuous operation was maintained be called by the other module. The likely tion. For simulating target tracking with a large number of attacking air- result would be for the common sub- algorithms, the target velocities and craft. All raids during this operational routine to lose its intermediate data as accelerations and radar measurement test were successfully dealt with. The well as its interrupt point. As mentioned noise are of primary importance. On the system is now undergoing shipboard earlier, the solution to this problem is other hand, target -related parameters testing aboard the USS Norton Sound on either to use reentrant subroutines or may be neglected in simulating the the Pacific Test Range for further evalua- make the modules nonpreemptable prior AN /SPY -1 control program for pur- tion.

www.americanradiohistory.com Toll switching plan for Alaska V. Batral E.F. McGill

In its three years of operation as Alaska's certified long -lines carrier, RCA Alaska Communications (RCA Alascom) has invested over 5100 million in plant and facilities to improve the state's commercial long -lines network. During 1973, RCA Alascom completed over eight -million long- distance calls - and with a growth rate of 25 to 30 percent per year. Alaska represents 1/5 the size of contiguous United States. It is important that a long -range plan be formulated to provide the most efficient service at the least cost. The toll- switching plan described here is a step in this direction.

A team of technical consultants in cooperation with RCA Alascom a modified plan, portions of RCA ALASCOM took over operation installed in Anchorage, and in May 1972 developed which are discussed in this article. of the commercial long -lines network of another such system was placed in service the Alaska Communications System in Fairbanks. (ACS) from the Air Force in January, Objectives of the switching 1971. In purchasing this system, RCA Customer reaction to DDD in plan Alascom set its sights on creating a Anchorage and Fairbanks toll centers. dynamic new communications system for which together handle about 80 percent of the modified plan were: the State of Alaska. All long- distance of the state's long distance traffic, was far The objectives calls, prior to take over, were handled via beyond expectation and a series of rapid Western Electric 3CL cord -type operator additions was initiated to meet the traffic a)To provide a backbone toll- switching to provide the best and most up -to- RCA Alascom's immediate volume. Maintenance during these network boards. date DDD service consistent with good priority was to establish direct- distance additions was difficult and in some cases management practices (i.e. the best service dialing (DDD), comparable to that the service suffered. By August 1972, at the least cost) for all areas of Alaska. provided in the contiguous United States. based on this new experience in DDD, a b) To stimulate additional revenue, not only by rates but also By February 1972, a North Electric NX- decision was made to re- evaluate growth increasing completion through influencing customer's calling 1 D with NT -500 ticketing system was projections. habits. c)To provide for growth capability in the Vinod Batra, Project Engineer. RCA Alascom, received E.F. McGill, Mgr., Engineering Planning, RCA Alascom. range of 20% per year up to 1990. the B. Tech in electrical engineering from the Indian received his AA degree from Clark College. and his BSEE Institute of Technology, Madras, India in 1964 and his degree from the University of Washington. He was the d) To provide for additional customer service M.S. degree from the University of Ottawa, Canada in Central Office Superintendent providing central office e.g. wide area telephone service. inter- 1967. From 1967 to 1969, he was with the R&D Labs, and transmission planning and operational direction for national direct- distance dialing and other Northern Electric Company where he worked on various Pacific Power and Light Telephone Company for over a functions required in Alaska's corn - aspects of crossbar and electronic switching systems. period of ten years. Mr. McGill became Switching municaitons needs. From 1969 to 1971 he worked at ITT -Canada, as a project Engineering Manager for RCA Alaska Commurications e) To meet and improve established engineer on the Mallard, all- digital switching system. in 1973 and Planning Engineering Manager in 1974. nationwide DDD service standards From 1971 to 1973, he worked on various aspects of toll - throughout Alaska. ticketing and switching design at Stromberg-Carlson in f) To provide capability to accomodate future He is presently the project Rochester, New York. technical developments e.g. satellite engineer for the new principal switching center at Reprint RE- 20-1 -3 transmission and switching. Anchorage. Alaska. Manuscript received March 4, 1974.

Vinod Batra (left) and E.F. McGill. Development of the switching plan

Fig. I represents the Alaskan long -lines network as it exists today. There are four class -4 toll centers: Anchorage, Fair- banks. Juneau, and Ketchikan. Each toll center provides a high usage and final route to the outside world. Presently, the control switching point for Alaska (907 area code) is Seattle. Calls travelling thousands of miles from the lower 48 have only two chances to complete in this last link. AT &T studies show that over half of such incompleted calls are never attempted a second time. Besides loss of immediate revenue. the future calling patterns are affected.

Based on these observations and objectives, the modified plan established the

www.americanradiohistory.com tensions. Above all, it should be economical.

FAIRBANKS Implementation of the CLASS 4 DUO switching plan NU o ANCHORAGE SEATTLE CLASS 4 DOD Implementation of the Switching Plan NU consisted of three major milestones: JUNEAU .-FINAL TO S TTLE CLASS 4 MANUAL 6b I) Study of present machine capabilities. KETCHIKAN 2) Selection of switching machines, and CLASS 4 MANUAL 3) Installation and cut -over of switching NU machines. TO SEATTLE

Fig. 1 - Toll switching in Alaska 1973. The switching machines presently operating at Juneau and Ketchikan are following priorities (Fig. 2): Anchorage. This switch will function Western Electric step -by -step, 2 -wire primarily as a control switching point for systems. They cannot be adapted to a) Establish DDD in Southeast Alaska. the 907 area code and will work in provide the growth, features, and re- 2 quirements of the statewide DDD b) Establish a control switching point (CSP) conjunction with the existing -wire NX- for the 907 area code in Alaska. I D, NT -500 system in a two -office (toll - network. Therefore, a new machine is c) Establish guidelines for the orderly growth center/ primary- center) configuration. required. of the switching machines to meet growth projections. Studies showed that Ketchikan is not Anchorage toll center on the other hand, d) Improve transmission on long -haul circuits large enough to economically support a is a modern 2 -wire cross -bar machine. It through use of 4 -wire, end -to -end switching center. In Phase could, with sufficient capital, be modified and transmission, if possible. toll III Ketchikan will cease to be a toll center and all tributaries to 4 -wire switching but is deficient in To establish DDD in Southeast Alaska, it will be homed on Lena Point switch - meeting the new service demands and the was agreed that a new switching machine center via a new microwave system. control switching-point functions. It was would be required. It was decided to therefore decided to freeze the growth of locate this switch -center at Lena Point, Phase IV of the switching plan addresses this machine and investigate switching approximately eight miles from Juneau, the Fairbanks Toll Center. A switch machines capable of meeting all re- will be made the capital city. Lena Point is the addition to the existing quirements. transmission hub of Southeast Alaska, equipment to meet the 1975 busy- season traffic forecast. Future growth direction thus an obvious choice of location. The selection of switching machines was will be studied as more data becomes However, Lena Point is far from being a accomplished, using twenty -year available. population center, hence a traffic forecasts on traffic, revenue, and costs operating system would have to be inserted into a present worth of annual During the development phase of the remoted from this switch to the town of charges computer program. Competitive plan, it became evident that any new were Juneau, the site of existing 3CL, cord - bids evaluated on the basis of both switching machine should be a 4 -wire type operating boards. the present worth of annual charges system with a high degree of flexibility (PWAC), technical requirements and ul- in built (which in turn implied a stored - timate growth capability. Based on this Anchorage, due to its central location and program control) and should have the study, Northern Electric's "SP -I, 4 -wire accessibility was selected as the site for the ability to be used at any level of the DDD ESS" distributed by Northern Telecom new primary center. The existing Network hierarchy as the Alaskan was selected for Lena Point and switching machine at Anchorage was network develops. The system must have Anchorage. found to be deficient in meeting the CSP low and easy maintenance, a high degree requirements. In Phase therefore, a a II, of reliability, and short installation The SP -I, 4 -wire ESS (Fig. 3) is a stored be second switch- center will installed at time for initial set up and future ex- program machine using a minibar switching matrix and a duplicated central processor. The solid -state, wired -logic and plug -in connectorized modules permit a compact package that is easily specified for initial installations and any FAIRBANKS CLASS 4 extension thereafter. This machine is fully NU compatible with the North American ANCHORAGE switching network, is designed for Class CLASS 3 LENA PT. CLASS car 4 I , 2, 3, or 4 applications, and will accom- JUNEAU TRAFFIC OPERATORS modate the Alaskan network as it con- 1 f tinues to expand. FINAL KETCHIKAN TO SEATTLE NU TOLL .tt , TERMINAL The machine features up to 10 -digit translation, code conversion, digit dele- Fig. 2 - Toll switching In Alaska 1975 - all DDD. tion, digit prefixing, unlimited alternate

www.americanradiohistory.com TOPS POST TION I - CIRCUIT INC downtown Juneau. An inservice date of TRUNK March 1975 is planned.

x The Anchorage primary center will be OTHER FROM N T ST INC 10 Oc iOFFICES OTHER [IP housed in a new three -story, 60,000 sq. ft., TRUNK Pu OFFICES TRUNK TEST LINK NETWORK air- conditioned building. Foundations TEST have been provided for additional floors, TRUNK CONTROL as future growth demands. The building r- TEST J will house switching equipment, a 0 REC TRUNK TOPS R TEST N POSIT ON LINN network management center, a traffic-

MANUAL operating system and administrative and RSLC TRUNK X TNLC TEST support facilities. The initial switch in- TO DISTRIBUTOR POSITION POINTS stallation will consist of 1,920 incoming,

REC MARNER SENDER MARKER 1,920 outgoing trunk appearances and 39 traffic operating positions. An inservice date of February 1976 is scheduled. Work on both switch centers in approximately 90% complete. CENTRAL CONTROL MAINTENANCE CENTER

Conclusion Fig. 3 - SP -1, 4 -wire toll- switching system with TOPS The population of Alaska has increased by about 40% since 1960. Alaska represents about one -fifth the total area routing, dynamic overload control, arranged in functional groupings and of the continental United States. This traffic and plant measurements, routine color coded. The view screen can be growth in population, linked with and demand testing - and a whole range adjusted in the horizontal and vertical Alaska's untapped resources, reflects the of other features which may be required plane to allow each operator to work vital need for a superior communication in the future. Each SP -I, 4 -wire ESS has a comfortably. system. During 1972, RCA Alascom maximum traffic capacity of 100,000 ccs completed 7- million long- distance calls. (call seconds) and the machine can be Besides the regular TOPS positions, aux- Current projections call for 35- million used in a multi -unit configuration to meet iliary positions will be provided to assist originated calls in 1980. The new Alaska's network demands through the operators, monitor operators, restrict switching configuration will help RCA 1980's. types of calls to certain positions and Alascom to meet this growth by in- receive force administration data. The troducing principal city functions in The traffic operating position system operators will be able to dial trouble Anchorage, provide direct-distance dial- (TOPS) will be introduced as part of the codes for various types of incompleted ing in Southeast Alaska, provide 4 -wire dialed or assisted SP -1 ESS package. Automation of subscriber operator switched transmission, provide a operator functions and subsequent reduc- calls. This data will be compiled, analyzed futuristic traffic operating position to the RCA DDD Service tion in work effort has always been a and outputted system, introduce standard toll -dialing prime goal of telephone administrations. Bureau. All billing information will be throughout the State of Alaska, in- on a magnetic tape and Introduction of direct- distance dialing recorded troduce the dial rate in the State of transmitted to the data center for sub- and the elimination of manual tickets are Alaska, introduce international direct - scriber billing. but two steps in this direction. In addi- distance dialing, and provide a com- tion, improvement in the working munication system equal to in capabilities All TOPS hardware will be maintained or greater than those provided by other environment of telephone operators is via the standard SP -I, 4 -wire techniques systems in the rest of the North American becoming more important both for job of fault detection. The diagnostic tests Continent. satisfaction and improved efficiency. SP- will be performed on a demand basis as I with TOPS will allow this to be realized well as on a scheduled periodic basis. for the first time in Southeast Alaska. The Acknowledgment fully stored program capabilities of the Switch status SP -I system will be used in the incorpora- The authors wish to acknowledge several tion of the TOPS System. members of RCA Alascom engineering, Lena Point switch will be housed in a operation and management staff for new 14,000 sq. ft. air conditioned, single - TOPS positions will consist of voice valuable assistance toward the develop- Besides the switching access to the SP -I, 4 -wire trunk network story building. ment of this plan. plus a video -display computer terminal equipment, the building will house the tied directly to the central processor. All satellite earth -station, power plant, ad- pertinent details of the call are displayed ministrative, and support facilities. The Bibliography on the operator's screen for the duration initial switch installation will consist of of the involvement in setting up a 1200 incoming and 1200 outgoing trunk Mc(;ill. E.. F. and Barra. Vinod.; RCA Alaska Com- munications. Inc.. "roll Switching Plan For Alaska." particular call. The operator can modify, appearances and 38 traffic -operating l'resentations to Alaska Telephone Convention. -I nrhorage, .Alaska, August 1973 and Alaska independent delete or add information through a positions. The traffic operating system lnatecting rompanie.H on an individual basis throughout teletype keyboard access. The keys will be will be remoted to the RCA building in the year of 1973.

www.americanradiohistory.com picture content will modulate downward, TTUE -4A uhf television exciter producing lesser power output. J.B. Bullock A positive- and -negative clipper follows the output of the final video amplifier to limit video peaks which might otherwise The TTUE -4A is the exciter now used to drive all current transmitters in the RCA uhf overmodulate. transmitter product line. It produces, from temperature -compensated crystal -oscillator sources, an amplitude -modulated visual carrier with much of the lower sideband removed, An adjustable unclamped reference level and, 4.5 MHz above that, a linearly modulated fm aural carrier. In the visual channel. the is provided via the clamp on off switch. In modulation is predistorted to compensate for phase and amplitude nonlinearities that will the off position, this adjustable reference inevitably take place in the power -amplifier stages to follow. In the aural channel, a choice level provides a means of setting the video of 50 or 75 microseconds pre- emphasis is provided, to be used as required in the given- signal and the resulting i.f. carrier to any broadcast system. desired point in their operating regions for test purposes.

Linearity correction VISUAL and aural modulation are and to the clamp -pulse generator. accomplished at intermediate frequen- The rf output of the transmitter (which cies. These i.f. signals, at 45.75 and 50.25 The clamp pulse is centered in time within usually employs a klystron power MHz respectively, are heterodyned the "back porch" at blanking level follow- amplifier) must be a linear reproduction against a common "pump" signal and ing each horizontal sync pulse. The clamp of the exciter's video input. Fig. 2 shows a pulse activates a sample- and -hold circuit "up-converted" to the desired uhf tv- typical saturation (power output vs drive which samples the voltage level of the channel frequency. Channel frequency is power) curve for a VA -890 type klystron. back porch at the input to the video thus dependent on only pump frequency, This not.- linear performance as satura- which makes much the exciter in- amplifier preceding the modulator and of tion is approached is typical of all power closes, for the moment, a high -gain dependent of the tv channel and thus klystrons. expedites both production and the stock- negative feedback path to the input of the This 35 -kW klystron is used only up to ing of spare parts. differential gain driver. This restores the the 32 level, which is to say that the rf back porch to its preset level and es- -kW power output at peak of sync will be 32 tablishes a blanking level at the input to kW. The transfer characteristic of Fig. 2 Visual signal the video amplifier which will be main- is replotted in Fig. 3 on a voltage basis tained by the clamp, independent of with 100% output voltage representing The generation of the visual signal is picture or sync amplitude. This clamped the 32 -kW output. It is seen from this indicated in the functional block diagram level is variable by an adjustable dc curve that at blanking (back -porch) level, of Fig. 1. Each block on the diagram inserted into the feedback loop in the where the output voltage must be 75% of generally represents a plug -in "baby" clamp amplifier. From the point of in the exciter. that at peak of sync, the input drive to the board clamping forward, all video circuitry is dc klystron must be 59% of the drive at sync coupled (or clamping would be lost). The incoming video signal is looped peak. Similarly, the drive at mid - through a differential input amplifier to characteristic (midway between blanking provide for cancellation of common - At the modulator, the back -porch level and reference white level) must be 33% of mode inputs. The video signal then goes becomes a reference i.f. power level from the drive at sync peak.

to the motor-driven (for front -panel which sync will modulate upward, Reprint RE- 20 -1 -20 remote operation) video level control, producing greater power output, and Final manuscript received August 13, 1973.

Stirs PHASE DA VER 45.T MN. 1.0 V VIS IF P P 0 VIDEO NVENTER oiFF DIFF eaw I N OF SYNC GAIN PIASE W CLIPPER CORR CORR BACK PORCH AT ZERO VOLTS IF 3GB POWER AMPS MOD IF LEVEL LEVEL CORR VIS OTIVE MOD FILTER CORR OFF

IF MOD SYNC POWER BIAS GAIN ..'- AMP BACK PORCH LEVEL (CLAMPED)

BUFFER

IF DRIVE LEVEL

VIS TOM t. AB.TB NHL

visual i.f. diagram. Fig. 1 - TTUE -4A exciter video - functional block 68

www.americanradiohistory.com a

100 3511W

100 91.5 32KW

87 80 o D.

á 1- 65.6 60

43.7 40

John B. Bullock UHF -TV Broadcast Engineering, Corn- munications Systems Division, Meadowlands, Pa. received the EE from North Carolina State University. He served in the U. S. Army Signal Corps during WW I fO 60 80 100 % for SATURATION principally as an instructor at the Army Electronics N 100 % for 32KW OUTPUT Training Center at Harvard University, emerging with the DRIVE POWER rank of Captain. Mr. Bullock has taken graduate studies at Yale and at the University of Pennsylvania. Joining Fig. 2 - Typical saturation curve for a VA -890 klystron. RCA in 1952, he became a mainstay in RCA's Microwave Relay activity in Camden and played a major part in system and circuit design of the TVM -1/3 and TVM -6/13 tv microwave relay equipment. In 1970, with the combin- PK of SYNC ing of tv relay and uhf transmitter engineering activities, he transferred to the Meadowlands plant where he participated in the final design stages of the TTUE -4A uhf tv exciter. Mr. Bullock was chairman of the BLANKING 75% EIA committee on tv relays from 1960 through 1970, and 67.5% is a member of the IEEE and a licensed professional engineer in the State of New Jersey.

MIO 43.7%

If the required klystron rf input 22.5% amplitudes of Fig. 3 are plotted against MAX DEPTH of the video signal to be transmitted, the MOD 12.5% necessary exciter transfer characteristic 20 40 60 80 100 results, shown in Fig. 4. (This RF INPUT VOLTAGE % characteristic excludes the ac- coupled (EXCITER OUTPUT VOLTAGE) stages at the exciter input, where the Fig. 3 - VA 890 klystron voltage transfer characteristic. video signal is inverted and its level set).

The exciter's rf output with staircase video modulation, rectified by a diode, and observed on an oscilloscope, will 4v appear as shown in Fig. 5b. Step risers 4 14V la 50, and subcarrier amplitudes here would be noted as up to 20% greater at blanking level, tapering to a constant value as reference white is approached. The 3.58 - !NANKIN MHz subcarrier in Figs. 5b and 5c is approximately half -amplitude due to most elimination of of the lower sideband MD at higher frequencies. At the modulator stage the signal is double sideband; hence modulation with the signal shown does ITE reach reference white level.

Fig. 4 shows that the only obviously 0 required non -linearity predistortion is sync and black stretch; however, white stretch and white compression are also provided in the exciter to perfect system

Fig. 4 - Required exciter voltage transfer characteristic. 69

www.americanradiohistory.com linearity. Some of the white stretch is in the voltage divider of Fig. 6. stretch (or both) as determined by adjust- required in improving the exciter's own Supplementary sync stretch is also ment of diode bias. Similarly, the ter- modulation linearity, which tends provided just prior to the modulator minating portion of the divider is shunted towards compression at modulation stage. via gated diodes to produce white com- depths. pression. The voltage- dropping portion of the The linearity correction, or predistortion, divider is shunted via gated diodes to is principally accomplished as indicated produce sync/ black stretch or white Four black- stretch shunts are provided, with gating bias adjustments starting at the tip of sync. Similarly, there are six white -stretch adjustments and four VIDEO INPUT VOLTS MT. white- compression adjustments. All ad- (a) justments simply set gating bias levels.

A transfer characteristic for the entire exciter is plotted in Fig. 7. Since the correctors are set to optimize linearity for the entire transmitter system, some predistortion is required from all the (b) correctors, including an approximate 2:1 sync stretch. This transfer characteristic curve, if replotted on a percentage basis, should closely coincide with the curve of Fig. 4 between reference white and tip of sync.

Just as the above compensating distortion is required to correct expected gain changes with level (differential gain), so a compensating distortion is required to correct for phase changes with level tor differential phase) to provide overall phase linearity, without which the video 5 modulation Fig. - Staircase waveforms. waveform would be distorted in the time domain. This compensation is provided by the differential phase correctors illustrated in Fig. 8. .2V

Differential phase correction is accomplished by gating, into the video line, an out -of -phase video signal at the level 475 n where predistortion is required. In Fig. 8, VID 2 is main video signal, while VID t z CR7 100, the -IV 1 and VID 3 are signals identical to VID 2 which are superimposed on dc levels of 6 - 20 and -6 V, respectively. This arrangement BLACK makes it possible to apply an adjustable +20 STRETCH CR3 ADJUST gating bias to the diode. When the diode RI is gated on, the opposite -phase video signal is added to the video line through +20 the 7 pF capacitance, shifting the phase of WHITE video while the con- -20 the resultant diode ADJUST = CR8 ducts. An approximate phase shift of 3° is +20-- -. obtained at 3.58 MHz. This provides a be to RI5 differential phase that can adjusted cancel differential phase of opposite sense WHITE 20 COMPRESSION - occuring elsewhere in the system. ADJUST This type of phase- correction circuitry has the advantage that virtually no Fig. 6 - Linearity (differential gain) predistortion. resistive loading is added to the video

www.americanradiohistory.com ®¡ iY::::::N:...... °. VID I

6.2 ,n..n. ..n..s.N.. r ..

I.NNN.. =...... :.:: VID 2 aii= a.mmnN nnn.., .nnmilñ: .. nn...... r, = B: o _ 221 11:18.=.SR W jrÌ'RÄrR.:

7PF

VID 3 2 76" j IOK" OIFF 2.2K" PHASE ADJUST

6.2

CR REVERSIBLE -VID

rw CR CONDUCTS, ADDING dd WHEN ANODE POTENTIAL OF DIODE IS AROVE CATHODE

Fig. 7 Exciter voltage transfer characteristic, clamped Input to rf Fig. 8 - Differential phase correction. output. line, thus the phase correction may be set Modulation video -level control sets the amplitude of without affecting differential gain. the video drive to the final video amplifier. The output of this amplifier, Seven sets of differential phase correction Following the corrector circuitry, a plus an added (adjustable) dc bias voltage are provided, with a phase- correcting reference test point is provided, and at drives the modulator. (predistorting) capability of about this point the "back porch ", or blanking ± 20 °. A phase correction in the white level of the video signal is adjusted to a dc region may be shifted to the black region zero voltage level (Fig. 1). This level is The modulator is illustrated in Fig. 9 by reversing the gated diode. maintained by the clamp circuitry. A together with a portion of the final video

.20V +12V

MODULATING AMPLIFIER

332" 475/1 40673 M 2N4034

220" 22" #50" 825"

390" 7F 7F 274" VIDEO 263866 - 50" INPUT DOUBLE SIDEBAND IOO" MODULATED V v IF OUTPUT = 1800" 1100" 27^

t CR4 20V 20V HP50B2-2800 MODULATOR MOD BIAS SYNC GAIN I IK" 2200^

BOARD I LIN ( 32K" IK"

I 150" SYNC tJ M :2200" CARRIER ADJ 2113947 INPUT 10K" 45 75MHZ 4Ó0,1W @KLIM,'AIN

-20V I -20V

Fig. 9 - Visual modulation. 71

www.americanradiohistory.com from the differential -gain- corrector cir- board in the exciter. cuitry. The carrier originates with a 418.75 -kHz temperature- compensated crystal oscillator. This frequency is multiplied by Sideband filtering 120 after modulation to reach the aural i.f. of 50.25 MHz and to provide a deviation capability of better than ±50 I he modulator produces a double - sideband amplitude -modulated signal. kHz at that frequency. The signal is then fed to an active Separate baseband audio and SCA Fig. 10 - Active filter response (spectrum analyzer bandpass filter where much of the lower display). Horizontal scale: 2- MHz /division; Vertical sideband is removed. The swept response (Subsidiary Communications, such as scale: linear. inputs are The of the active filter is indicated in Fig. 10. telemetering) provided. former passes through a choice of 50 or 75 Note that the lower 3.58 -MHz sideband is amplifier and the motor- driven sync gain attenuated about 6dB, leaving the bulk of as pre -emphasis and then both are fed to control. The latter adjustment can in- a series of five serrasoid modulators. In sideband shaping to be done by the crease the gain of the modulating sideband filter following the power this type of modulation, each modulator amplifier in the sync region by reducing produces a pulse whose is amplifier in the transmitter. time position the negative feedback around the advanced or retarded from the pulse at its amplifier to provide additional sync input in to modulating Following the active filter, the visual i.f. proportion the stretch. About 40% of the stretch usually voltage at the time of its occurrence. Each signal is taken through a series of three required in operation is obtained from modulator operates on the output pulses power amplifiers ( Fig. 1). These bandpass this control. The feedback is reduced of the preceding modulator; thus the total units are adjusted to be flat from 44 to 51 when bias to the CR4 diode allows it to be phase displacement is MHz leaving band shaping in control of produced the sum gated on as the video signal nears of the five sync the active filter. The i.f. signal is raised to displacements produced by the level. a level of 2 watts at peak of sync by these modulators. The resulting signal would be a stages, and is then fed to the visual phase modulation of the carrier. Modulation of the 45.75-M Hz visual i.f. upconverter. However, as is well known, a frequency - signal is accomplished by the FET- modulated signal can be produced by a transistor combination: the video signal phase -modulation process if the on gate I of the FET varying the gate -2 Aural signal modulating signal is retarded 90° and transconductance. In operation, the gate - attenuated in proportion to its frequency

I (MOD) bias is set such that the full Origination of the aural i.f. signal is prior to modulation. A network for amplitude of sync is preserved, allowing indicated in the functional block diagram accomplishing this is incorporated with

some white compression, which is then of Fig. I I. As in the visual portion each the pre- emphasis in the processing cir- corrected by the white stretch available block generally indicates a plug -in "baby" cuitry preceding the modulators.

SCA INPUT fNET

SCA DEV 1 T

600/ 150n NET +10DBM dAUD AUDIO C DEV INPUT LF DIST ADJ rv +20

PRE -EMP AURAL PULSE MOD MOD MOD MOD D TCXO SHAPER T418 KHZ IUNIVIBRATOR HF DIST ADJ - (SO WAVE SYMMETRY)

IINTEGRATOR

50.25MHZ x.. AURAL IF 2 w X 5 .v. TO . 1x2 t lC.. X3 t 4P 3d8 UPCONVERTER 418.75KHZ 837.5KHZ 167 MHL 3 35MHZ 10 05MHZ / \ IF 20041W MULT MULT MULT CRYSTAL MULT CRYSTAL MULT LEVEL BUFFER IF BANDPASS BANDPASS POWER vii TFR FILTER AMP

Fig 1 - TTUE -4A exciter audio - aural i.f. functional block diagram.

www.americanradiohistory.com MI (FIGURE AURAI 30 25MMZ UPCONVERTER 200416 646 475- 606MM2 AURAL IF

212 361MMZ

75 +6dB PIN DIODE ATTELA

32 CM 14-36 I 6W 9 PDW 33 CM 37-69 HORRID ---4CwPLER 21 7 SENS 21W -346 - 2 ese

3 2W 0e P66 LIMITER 4 46

F17 5 4646 5W

2646 CIRCULATOR 3m.

95V TClfO 14W 7-17 L0 TO ISv MMZ (FIGURE 11 VISUAL 43.75YNZ 414 PR OF SIRIO 1MCORV[R7ER IV/ PR OF SYNC 471-602MNZ VISUAL IF 641

Fig. 12 - TTUE -4A pump power rl generation block diagram.

Frequency multipliers of a phase shift of nearly ±2 radians. so final pump frequency at 527.5 MHz). that the five cascaded modulators can The modulated signal at 418.75 kHz, at produce about a phase shift of ±10 The three 6 -dB power amplifiers (PA) are the output of the last modulator, is a radians. At a modulating frequency of 50 identical units. The two paralleled units series of positive pulses with the modula- Hz, this amounts to a frequency change are matched in power output by upsetting tion information in the position of the of ±500 Hz from the 418.75 kHz fun- the input VSWR of the higher -gain unit leading edge. These pulses then key a damental. The multiplication by 120 via a small trimmer capacitor. The monostable multivibrator to produce a raises this deviation to ±60 kHz, ade- matching is done to minimize losses to the square wave that is integrated to produce quate for European standards. Operation adding hybrid reject port, since available a back -to -back sawtooth waveform, still to US standards ( ±25 kHz) is possible pump power is limited by the dissipation at 418.75 kHz. The multiplication process using only three modulators but with a ability of the transistors in these then begins by applying the back -to -back minor increase in distortion. paralleled PAs. A power limiting adjust- sawtooth to a series of three full -wave ment in the 14-dB PA is set to limit the rectifiers, each preceded by an amplifier drive to the higher of the paralleled PAs with a nominal 2:1 gain to restore the Pump -power generation to that producing a 22 -W output. This amplitude. Each amplifier, plus full-wave avoids PA transistor burn -out in event of rectifier, constitutes a frequency doubler. The visual and aural i.f. signals created at an automatic level control (ALC) with no tuning adjustments being 45.75 and 50.25 MHz, respectively, are malfunction. required' Symmetry of the driving hetrodyned up to any desired channel in square wave is carefully maintained to the 470 thru 806 MHz frequency range assure the symmetry of the back -to -back (channels 14 thru 69) in two separate The adding hybrid following the PAs is ,awtooth waveforms. Departure from upconverters driven from the common identical to the splitting hybrid but used this symmetry introduces distortion, and pump source. The generation of pump "in reverse." the symmetry adjustment is used as a power is illustrated in Fig. 12. distortion nulling control. An ALC loop surrounds the amplifiers The pump signal originates with a TCXO and multiplier following the X8 mul- A tripler employing crystal filters at input in the I I.7 -to 17.4-M Hz frequency range. tiplier. This loop permits setting the and output, and a harmonic multiplier A specified TCXO is used for each of the pump -power level and acts to maintain tuned to the fifth harmonic complete the uhf channel frequencies including those the level of power out ofthe power sensor multiplier chain, raising the signal to the requiring 10kHz or -10kHz offsets. For constant at the required 21 W. The ALC final aural i.f. of 50.25 MHz. A tuned channels between 14 and 36, the total is needed primarily to compensate for buffer amplifier and broadband power multiplication is 32; for channels 37 thru changes in amplifier gains with amplifier then provide the required 69, the factor is 48. Typical power levels temperature. power output of 400 mW. These two at each stage in the chain are shown on amplifiers are identical to stages used in the block diagram. The final pump signal Incorporated in the ALC amplifier is an raising the visual TCXO output at 45.75 has a 2nd harmonic about 55 dB below input which will reduce the output of the MHz to a 400-mW level prior to the the fundamental, and no close -in signal ALC circuit to zero upon application of a visual modulator (Fig. I ). higher than 70 dB down. (This data was -2 V to that input. Zero output trom the taken from a pump chain at channel 31, ALC amplifier causes maximum Each of the aural modulators is capable TXCO frequency at 16.484375 MHz, and attenuation in the pin -diode attenuator

www.americanradiohistory.com following the X8 multiplier and thus reduces the pump -power output to zero. This feature is used to remove drive CRO power from the transmitter klystrons in 20dB VIDEO -20dB (see FIq.IS) the event of a vswr fault in the line INPUT output DIODE VISUAL 16dB L75n of any klystron. The vswr fault circuitry OUTPUT t-20d8

furnishes the -2 V to the ALC. This 75 TTUE -4A AURAL 510dB feature saves the klystron and its output EXCITER OUTPUT 50^ BWU -5 line from further damage ifa VSWR fault DOWN should occur. CONVERTER CR0

VIDEO Is. Fia. 151 SWEEP V SYNC a BW -5 Upconverters BLANKING SIDEBAND ADDER ANALYZER The visual upconverter and aural upconverter are identical units, however, Fig. 14 - Exciter (visual upconverter) final broadbanding response test block diagram. the alignment of the former is much more painstaking in view of its linearity and bandwidth requirements. of the varactor junction capacitance generated by the upconverter itself, and The upconverters use a Varian VAB802 affects tuning of both the pump tank and that no added dc be drawn from the varactor diode in the circuit arrangement the first section of the output filter, and to varactor bias supply due to the presence shown in Fig. 13. All inputs and outputs a lesser degree, the i.f. input impedance. of i.f. drive. are at 50 ohms. Nominal maximum test

power levels are shown on the diagram. As the output filter is tuned, CR 1 must be Final stages of visual upconverter align- Gain of the upconverter from i.f. -in to rf- "worked into" the pump and filter circuits ment are done with the upconverter out is 6 dB, with a 7 -MHz passband. installed is by patient adjustment of the C7/ CR 1 and in the exciter. The exciter then C8 /CRI ratios; the pump tank must be modulated by the video sweep output of a Initial alignment is done using a swept 44- kept resonated; and the i.f. and pump sideband analyzer with sync and blanking to 51 -MHz i.f. input, with a pump power inputs must be matched to 50 ohms. added. Using this arrangement, as shown of about 7 W. For alignment, the in Fig. 14, the power level at peak of sync reflected pump and i.f. -input powers When an rf output of 2 W has been can be held at 4 W by observing the diode; must be minimized and power output at achieved at the 7 W pump level, pump unwanted spectrum components can be the pump +i.f. frequency be maximized. power is raised to 14 W and alignment is held at proper levels by observing the Initial alignment is done at reduced pump continued until the 4:1 -gain, the 4 -W spectrum analyzer; and the desired pass - power to avoid developing excessively output, and 7 -MHz bandwidth are band achieved by observing the sideband high rf voltage across the (initially) un- achieved with no more than 0.5 dB of analyzer. The alignment will be con- loaded pump tank circuit. compression at the 4 -W output level. sidered complete when the swept response does not "tilt" by more than It is easily seen from the circuit of Fig. 13 Alignment to this point also includes the ±0.5 dB as picture brightness is varied that there will be much interdependence requirement that the pump signal at the rf through those levels which would in the various tuning adjustments. The output be at least 30 dB below the 4 -W produce 22% thru 67% of final varactor bias which sets the initial value level, that no spurious signals are transmitter -output voltage. (Referring to

FILTER 30V LI L2 RI I L3 VIIF AT-I BIAS PA 3dB I W <,05W C2, C3 L4

PUMP CAVITY TTPUT FILTER

INPUT) LS CS C7, I XCB 7 VIS 7j» ALS í C6 C9 L6 CIO CII CIRC J2 .1 J3 14W CRI 4W. «2W (51 4W OUT (IW IF IN) .< 6W ®MIN OUT (MIN ZF IN) J

VARACTOR EQUIVALENT VAB -602

Fig. 13 - Upconverter - simplified schematic with test metering.

www.americanradiohistory.com Fig. 3, this is seen to be a variation of which shows a view of the exciter with the from 17% thru 52% of the exciter's output visual i.f. drawer open. All drawers slide voltage at sync peak). forward in this manner. Each vertical circuit board seen in the figure carries The diode and sideband analyzer circuitry of one of the blocks in the responses obtained from a properly diagram of Fig. 1. The bottom row. for operating exciter upconverter) in the test example, from front to back contains the setup of Fig. 14 are shown in Fig. 15. The visual i.f. TCXO, buffer amp, and i.f. response shown on the sideband analyzer power amp. These vertical boards plug is virtually identical to that of the active into sockets on "mother" boards which filter (Fig. 10). The roll -off in diode are also vertical but run from front to rear response shows the elimination of the of their respective drawer. These mother lower sideband at higher modulation boards provide interconnection between frequencies. Tip of sync in this photo is at the "baby" boards via printed circuitry the 4 -W level. and also mate with the cable harness connector. The cable harness then Packaging provides interconnection between mother boards in this and the other The exciter is contained in one horizontal drawers. and four vertical drawers in the 28 -in.- The right -hand vertical Fig. 16 - TTUE -4A exciter photograph - visual i.t. high rack -mountable frame pictured in drawer in Fig. 16 drawer open. Fig. 16. The horizontal drawer at the top is the pump drawer. In it are mounted the of the frame contains metering and con- circuit stages illustrated in Fig. 12, star- of their output voltage, the programming trols. The four vertical drawers contain ting with the X8 multiplier. These stages, resistors which set the voltage (from left to right) the aural i.f., visual i.f., because of the frequency and power levels output being mounted on the video processing and low -level pump involved, are built in tightly shielded mother boards and wired via socket connections. Negative stages, and higher level pump chain plus chassis with heat radiating fins and / or regulators are also interchangeable. Both upconverters. heat sinks to the chassis of the drawer. types of regulators incorporate overload and overvoltage protection, thus protec- The exciter frame is mountable in the ting themselves from damage by load control cabinet of any of the current Metering fault and protecting their loads from product line of RCA uhf transmitters, or, damage by an overvoltage fault. alternatively it may be mounted in a Each stage in every area of the exciter has separate cabinet for service as a spare a dc metering lead brought to the front exciter. panel meter with a reading indicative of Acknowledgment normal operation. In this manner, trou- The circuit packaging in three of the ble can quickly be isolated to the stage The TTUE -4A uhrty exciter represents vertical drawers is as displayed in Fig. 16, involved. The tront metering pane, also several years of development effort by includes controls for setting video Meadowlands Television Broadcast modulation level and sync gain (stretch) Engineering personnel. Particular credit plus jacks for connection of a counter to is due 'I . M. Gluyas, D. Gray, and W. L. make direct measurement of visual, Behrend on the overall exciter systems OUTPUT aural, and pump crystal oscillator fre- and to R. E. VOLTS Desrochers for integration -0 quencies. into the uhf transmitter product line. Visual i.f. design was largely the work of -33 %(MIDI D. Gray and early upconverter design -59% Power supply credit is due W. L. Behrend. The video clamp and corrector circuitry was fhe exciter is powered by -100 %(SYNC TIP) 4W a separate dc developed by J. A. Bordner, the voltage supply which operates from 200 - audio/ aural circuits by D. Yerzleyand R.

1 1 0 4MHZ to 25OV, 50- to 60 -Hz power. Four dc Brankley, and final upconverter and DIODE DISPLAY voltages are supplied to the exciter as pump chain design was by W. M. Boyd. (NEGATIVE RECTIFIER) follows: +45 V, -25 V, -30 V for up- W. S. Day was responsible for converter bias, and ±24 V for control mechanical design and final packaging. motors. l'he author recently relocated from Camden, participated in the final stages of design and has been active in the In the exciter itself are eight positive factory and field followup program. regulators and two negative regulators. Helpful editorial assistance on this paper All are plug -in "baby" boards and each was provided by R.E. Wolf. provides a required voltage to some ó 4MHZ subdivision of the exciter. Three regulators may be seen in the upper Reference compartment of Fig. 16. All positive Fig. 15 - Diode and BW -5 sideband analyzer displays test setup of Fig. 14. regulators are interchangeable regardless I. Relerence LS patent 3596208. 1). L. Yrnlev

www.americanradiohistory.com Laser scanning as a tool for testing integrated circuits

Dr. R. Williams) P.V. Goedertierl Dr. J.D. Knox

Laser scanning can be used to test integrated circuits for defects. This paper describes the method and shows some results for a simple circuit.

GREATER RELIABILITY- in the to look at integrated circuits. Each manufacture of integrated circuits means functioning element is probed by se- that new testing methods must be in- quentially scanning the entire circuit with vented to keep up with advances in circuit the focused spot from a laser beam. This design and fabrication. At RCA is done at some stage of fabrication before Laboratories, we are exploring a new way the circuit is encapsulated. At those

Left to right: authors Knox, Williams, and Goedertier.

Peter V. Goedertier, Physical Electronics Research Dr. Richard Williams, Physical Electronics Research Laboratory, RCA Laboratories, Princeton, NJ obtained Laboratory, RCA Laboratories, Princeton, NJ, received the MA in 1941 from the University of Louvain, majoring the AB in chemistry from Miami University and the PhD in theoretical physics. He held various positions in in physical chemistry from Harvard University. He joined teaching and research field at the Philosophical College the laboratories in 1958. He has worked in several areas, of Louvain, The University of Louvain, and Duke Un- including semiconductor -electrolyte interfaces, liquid iversity. At RCA Laboratories, which he joined in 1959, he crystals, internal photoemission, and insulator physics. has been engaged in early work on optical masers, with He was group leader in Insulator Physics from 1967 to both gaseous and solid -state materials. He was the co- 1970 and in Quantum Electronics from 1970 to 1972, discoverer of the He-Ne "cascade" laser, and in 1965 was when he was made a Fellow. He has received several given an Achievement Award for his contributions to the achievement awards and shared in the David Sarnoff development of the cross -pumped YAG:Nd:Cr laser. For Team Award in Science in 1969. He was a visiting the last three years, he has been working on the scientist at the RCA Zurich laboratory in 1963 and has development of various optical systems, relating to the been a Fulbright Lecturer in Sao Carlos, Brazil, a summer processing and thedisplaying of video information. He is school lecturer at Instituto Politecnico Nacional in the author of more than a dozen scientific papers, and Mexico and a visiting lecturer at Princeton University. holds three U.S. patents.

www.americanradiohistory.com places where there is no lead metalliza- DEFLECTORS ARGON We chose the simple COS/ MOS circuit, tion, the light enters the underlying AND OPTICS LASER CD -4007, to demonstrate the method. silicon. Here it is absorbed to produce The CD -4007 is a dual complementary free holes and electrons. Wherever there TO VIDEO pair plus inverter, with three -n- channel AMPLIFIER is a pn junction, these move in opposite and three p- channel enhancement -type directions. This transient charge flow INTEGRATED YCPHOTOMULTIFLIER MOS transistors, connected as shown in gives a signal that can be detected at CIRCUIT o Fig. 2(a). The optical microscope picture various points in the circuit as the spot of Fig. 3 shows the layout on the chip. To passes over the junctions. We take out emphasize the symmetry of the we Fig. 1 - Schematic drawling of laser- scanning ap- circuit this signal and use it as a video input for paratus. The video signal can come either from the connected some of the leads together as circuit itself, or from the photomultiplier. display on a kinescope, which shows a shown by the dotted lines of Fig. 2(b). "picture" generated by the actual Leads 14 and 7 then go to the input of the functioning of the active circuit elements. video amplifier, and the circuit is scanned necessary, and A bad transistor gives itself away by its compress the scanned area with the laser beam. To bring out so that it just covers pathological response to the laser spot. the surface of an different effects, we can apply a small dc integrated This gives us a rapid test for circuit chip. To get a general voltage, + or to the input gate, terminal picture of the circuit by reflected light, we malfunctions: a direct visual indication of 6. Alternatively, we can leave it floating. the bad spot and where it is on the chip. use the apparatus simply as a flying -spot scanner: The reflected light is picked up by a photomultiplier and the We first use the scanner as a flying -spot The laser scanner' (shown in Fig. 1) uses a photocurrent is used as a video -input scanning microscope to produce a picture l -W argon laser (4880A) with a combina- signal. This gives a magnified picture by of the circuit by reflected light (Fig. 4). tion of deflectors, one acousto -optic and reflected light. The picture does not show This helps with optical alignment. Then the other a scanning- mirror galvano- defects that are not evident under an we take the electrical output generated by meter. These generate a 525 -line video - optical microscope, but it can be used to the circuit itself to generate pictures raster scan at tv frame rates. We can focus align the chip rapidly and to identify such as 5(a), 5(b), and 5(c). These are the the laser spot to a small size, 3µm if different parts of the circuit. kinescope displays when w use the cir-

14 2 I I 14

Dr. Joseph D. Knox, Physical Electronics Research Laboratory, RCA Laboratories, Princeton, NJ, received the BSEE from the University of Dayton in Dayton, Ohio, in 1966, the MSEE from Case Institute of Technology in Cleveland, Ohio, in January 1970 and the PhD also from Case Institute in September of 1970. In September of 1970 he joined RCA where he has taken part in the design IO 4 _10 12 and development of a laser- deflection display system where he has worked mainly on the development of a Q6 large aperture and compact optical system incorporated within the deflection system. To date, he has designed and developed three large- aperture compact optical systems. He has also submitted a number joint of patent 4 disclosures dealing with the applications of the laser address system. He is continuing his work with the further refinement of the optical systems. His other TERMINAL NO. 14 VDD TERMINAL NO. 14 VDD activities include the design and fabrication of acousto- TERMINAL NO. 7 TERMINAL NO. 7 = VSS VSS optic deflectors, modulators and cavity dumpers for visible and infrared lasers. The tellurium dioxide deflec- (a) tors and latest lead molybdate deflector incorporated (b) within the laser deflection system were of his design and Fig. 2 - (a) Circuit diagram for COS/PADS CD -4007. (b) Connections for displaying fabrication techniques. circuit output during laser scanning. The video signal generated by the circuit is taken from leads 7 and 14. A dc voltage may be applied to the input gate, 6, or it may be left floating.

Reprint RE- 20-1 -4 Final manuscript received December 17, 1973.

www.americanradiohistory.com Fig. 3 - Optical microscope photograph of circuit, COSiMOS CD -4007. The Fig. 4 - Kinescope display of circuit as viewed when the laser scanner is used as a flying three p- channel transistors are in the top row and the three n- channel spot scanner. The light reflected from the chip is picked up by a photomultiplier and the transistors are in the bottom row. resulting photocurrent used as the video input signal.

cuit Output as the video signal. They show but the scanner identified the problem dicates a had element visually; it is the responses of the n- channel and p- immediately. different from a good one. We believe channel transistors and how an input that the information from this new test voltage on the gate enhances one or the will be a useful addition to that from other of these. In a properly functioning How generally useful is this method going existing methods. circuit. all the n- channel transistors to be? The results will depend a lot on the behave alike and all the p- channel circuit. The test has to be done before transistors behave alike. When one of encapsulation. Any circuit will generate a Acknowledgment these is had. it is immediately obvious picture showing the functioning from the laser scan. For example, in Fig. elements. Effective testing will require We are indebted to A.C. Ipri, W. C. 6 the middle p- channel transistor is maximum use of the symmetry of the Schneider, and J.H. Scott. Jr., for visibly different. (The middle transistors circuit and of its layout on the chip. The valuable discussions and help with the on the chip are actually those at the left on easiest case is one where many elements experiments. the circuit diagram (Fig. 2); the Dad one is respond alike in a perfect circuit. You that connected to the terminals 13,6, and simply look for one that is different in the 14.) Tests on a circuit tracer showed that chip you are testing. What the laser Reference the gate insulation on this transistor was scanner does is give a new look at the

I (- g. I.: Knox. J. D.: and Goedertier. P. V.:"A Television faulty. Under the optical microscope, we a generated directly circuit "picture" Rate Laser Scanner I. General Considerations." RCA could see no defects in the bad transistor. by the functioning elements. This in- Rrrirrr Vol. 33, No. 4 114721 p. 623.

(a) (b) ( c)

Fig. 5 - Kinescope display generated by using the signal from the active circuit elements as the laser spot Fig. 6 - Kinescope display generated by a defective passes over the circuit. (a) input gate floating. (b) +3 volts on the input gate. (c) -3 volts on the input gate. circuit. The central p- channel transistor gives an A positive voltage on the input gate enhances the signal from the p- channel transistors and a negative anomalous response. Conventional testing showed that voltage enhances that from the n- channel transistors. Connections as in Fig. 2 (b). the gate insulator on this transistor was shorted out.

www.americanradiohistory.com Charge -coupled imager Vertical cell count The RS -170 vertical blanking interval is .075 ± .005 of the field interval yielding for 525 -line television 241.5 to 244.125 (243 nominal) active R.L. Rodgers, III display lines per field. This places a requirement of a minimum of 242 lines to be generated by the imager for each field.

The operation and performance parameters are described for a CCD image sensor having The maximum number of lines in each 512x320 elements (163,840 individual storage sites). The device is designed and fabricated field is 262 to enable all of the cells for that for use in the standard 525 -line television system. This solid -state imager is capable of field to be read out before the next field performance superior to that of camera tubes with regard to signal -to -noise ratio, freedom starts. One row of cells is required for from lag, and absence of microphonics. Additional features include its small size, light each line in the display. The RS -170 sync weight, low power, and precision image characteristics. pulses cause the display to interlace each field with the preceding field. This causes a display with approximately 486 active lines per frame. The same 243 line positions generated on one field may be CONSIDERABLE EFFORT went by the National Television System Com- repeated on the next field giving a 486 into the development of a tv system that mittee (NTSC). This standard is active line display, with the resolution of would make an acceptable picture utiliz- described in EIA standards RS -170 for 243 lines, or a new interlaced set of 243 ing all of the eye -brain visual perception broadcast and RS -330 for closed circuit elements may be supplied by the sensor principles. Factors such as the visual industrial applications. For the purpose acuity of the eye and critical flicker of the following discussion, the re- frequency effects were considered. This quirements are the same. R.L. Rodgers 111, Manager, CCD Manufacturing and Engineer ng, Electro -Optic Products, Lancaster, Pa.. effort led to the development of the U.S. received his training in physics and electrical engineer - Standard 525 -line television system. For The standard RS -170 system consists of ing at the Polytechnic Institute of Brooklyn. He has been working on electro -optic imaging devices since joining a two 262.5 line fields each 1 60 second solid -state image sensor to obtain its RCA at RCA Laboratories in 1964. He was instrumental in greatest market potential it should be long (I/ 59.939 second for color), in- the conception and design of RCA's Silicon Vidicon and Silicon Imensifier Target (SIT) camera tubes. He compatible with and meet the minimum terlaced 2: I to form one complete frame transferred to the Electro -Optic Products Advanced performance parameters of this standard every 1/ 30 second. Approximately 23% Technology area in Lancaster in 1969 to continue work system. Many approaches have been of each frame is devoted to blanking time on silicon camera tubes and LLLTV cameras using these taken toward achieving this goal. The for the display. This time is used to tubes, resulting in the development of a photoelectron noise limited I -SIT camera. He was responsible for the most promising approach thus far is a transfer charge from one register to development of an effective non -blooming silicon target charge- coupled imager which has the another in a CCD. design and new lower cost fabrication tecnniques for silicon targets- Most recently he has been involved with potential for meeting both performance Charge Coupled Devices (CCDs) for imaging and and manufacturing requirements. The Reprint RE- 20 -1 -1 memory applications. He was promoted to his present production experience with silicon Manuscript received April 24, 1974. This paper was position early in 1974. Mr. Rodgers has presented many presented at the 1974 IEEE Intercon in New York (March technical papers and received an RCA Laboratories vidicon targets will provide a base for the 1974). Achievement Award in 1968. manufacture of solid -state charge - coupled image sensors'.

A 512 X 320 element CCD imager containing 163,840 individual analog storage sites has been designed and fabricated for the requirements of the standard 525 -line television system. In this paper, the requirements for this system are defined and the resultant imager performance is described. Several smaller size vertical frame transfer CCD imagers were developed prior to the 512 X 320 imager "''4. These devices demonstrated that the required device performance parameters could be obtained.

Image sensor requirements for 525 -line tv compatibility

The television system used for almost all broadcast and closed circuit television in the United States is the system described

www.americanradiohistory.com (each field) for a total of 486 lines of movie film, and still photography. Most Table II - Popular image formats. film and tv formats have the 4:3 vertical resolution per frame. movie Movie Film Format Miag.) Aspect Ratio ratio which is required for the aspect 8 MM 5.46 MM 4:3 standard tv system. There has been a Super 8 MM 6.68MM 4:3 Horizontal cell count general trend towards smaller formats 16 MM 12.0 MM 4:3 35 MM 25.5 MM 4:3 The RS -170 standards call for a horizon- through the years in all three media. This tal line frequency of 15,750 lines / s (15,734 has been the result of improvements in the Television lines / s for color). The line time is resolution of lenses and photosurfaces as 2/3" Vidicon 11.0 MM 4:3 therefore 63.49 microseconds. The well as the need -for more compact and 1" Vidicon 15.9 MM 4:3 1.2" Vidicon 21.2 MM 4:3 is lightweight equipment. The most popular nominal horizontal blanking interval 1.5" Vidicon 25.0 MM 4:3 17% of the horizontal line time leaving lenses are the "C- mount" variety. There 52.7 microseconds active display time. In are two basic subdivisions of C -mount Still Film closed circuit applications, the horizontal lenses. These are the designs for the 16 110 21.2 MM 4:3 126 39.2MM 1:1 video bandwidth is unlimited. In mm format of one inch vidicons and 16 135 43.3 MM 3:2 monochrome and color broadcast, the mm movie film. Any of the 16 mm vidicon luminance bandwidth is 4.2 MHz. Filling lenses can however be used for the smaller this bandwidth requires approximately formats. The II -12 mm formats seem Spectral response 450 cells. A composite color signal has the very attractive because of their small size on a 3.58 and ready availability of inexpensive color information modulated The spectral response of a CCD is similar video interchangeable lenses. Lenses for 8 mm MHz subcarrier. Unless the to the spectral response of a silicon and Super 8 mm have low MTF's in the tv monitor or receiver uses an expensive vidicon. The spectral response extends filter to remove the chroma infor- line ranges of interest and the small comb from a .4µm to 1.1µm. Fig. 2 shows the luminance signal (not format requires extremely small CCD cell mation from the for a thinned area dimensions- beyond the state of the art spectral response commercially used), the color subcarrier imager test device. Absorption of light in with the luminance signal causing for fabrication and device operation con- beats the transfer electrodes reduces the siderations to achieve standard tv resolu- an annoying edge creep. It is therefore available light to the substrate when an common practice to limit the luminance tion (240 television lines per picture image is formed on the electrode side. video bandwidth to 3 MHz. Most height). An additional drawback of very When the light is incident on the backside present -day video tape recorders are also small formats is the inability to achieve (side opposite transfer electrodes), the limited to 3 MHz video bandwidths for selective focus special effects with normal deviation in sensivity from 100% quan- luminance information. This leads to a fields of view due to the inherent very tum efficiency is caused primarily by minimum number of cells for large depth of focus. practical reflection losses and incomplete absorp- the 3 MHz bandwidth of 320 cells. The tion in the infrared. The magnitude of video data rate is 6 MHz. corresponding light loss in electrode side imaging varies Table I summarizes the 525 -line system 512x320 CCD imager with process variations and wavelengths, requirements. performance parameters and can differ somewhat from that shown in Fig. 2, particularly in the blue end of A 512 X 320 element CCD array has been Picture format considerations the spectrum. The backside illumination designed and fabricated for the standard is basically independent of electrode 525 -line tv system as described above. It is There are several factors affecting the details. choice of picture format. Table II lists the capable of generating the full resolution most popular formats used in television, requirements of broadcast color receivers and tape recorders. Fig. 1 shows the SIT -CCD this vertical frame transfer Table I - Summary of 525 -line television system re- layout of quirements (RS -170). imager. The imager is composed of three A vertical frame -transfer CCD imager

Vertical Parameters , sections. The lower section contains 256 processed for backside illumination that X response may be fabricated Frame Time (1/30 Second) 320 cells which can be interlaced on has good blue 2 Interlaced Fields (1/60 Second Each) alternate fields to generate 512 X 320 into a new type of high-sensitivity CCD Total Number of Lines Per Field (262.5) picture elements per frames. In actual use, imager. This device is the CCD version of Vertical Blanking Interval (.075 t .005 V) only 486 lines (243 per field) are displayed the Silicon Intensifier Target (SIT) Active Scan Lines Per Field (Nominal /243) as described above. The extra elements Camera Tubeb. The SIT -CCD imager Required Vertical Cell Count: are provided to allow variations in system consists of a photoemissive cathode, 242 -262 Each Field (Repeated or Interlaced) blanking and timing and to avoid non - electron -optical focusing section, and uniformities at the picture edges. The CCD imager in an evacuated envelope. Horizontal Parameters middle section contains a 256 X 320 cell The photoelectron image is injected into Luminance Electrical Bandwidth 14.2 MHz) storage area to provide format conver- the silicon at high energy, forming many Color Receiver and VTR Practice (3 MHz Bandwidth) sion to a sequential horizontal readout. secondary hole -electron pairs. This Required Cell Count: The top section shows the 320 -element current gain greatly increases the 4.2 MHz (Approx. 450 Cells) readout register which shifts the video out sensitivity of a CCD by raising the output 3 MHz (Approx. 320 Cells) at 6 MHz data rate. The cell size is 1.2 X signal above background signal variation 1.2 mils resulting in a 12 -mm image limitations. Interline- transfer CCD im- Format Parameters format. The overall chip size is 500 mils X agers cannot use this method of low- light-

Picture Aspect Ratio 14:3) 750 mils. level sensitivity enhancement.

www.americanradiohistory.com Resolution responses depend on the phasing of the Table Ill - Resolution loss mechanisms. light pattern with respect to the cell Image Formation Losses The most basic limitation on limiting pattern. If the black- and -white bars are resolution is the number of cells con- lined up with the cells in an in -phase Mechanism 1/2 Cell Limit Cell Limit 12Cells/TVL) 11 Cell /TVL) tained in the imager. In the horizontal condition, the response is greatest. If the 111 Lens MTF 94 .88 direction, the number of cells is bars are lined up with half a white bar and (21 Sin XIX Cell Sampling determined by the number of channels half a black bar contained in each cell, the MTF 90 .63 iñl formed by the channel stop diffusions be zero. response at the cell limit will 131 Lateral Diffusion Ml F .95 90 (320 for the 512X320 device). In the MTF Product 1.2,3 .80 .50 vertical, the number of resolution Transient response elements is determined by the number of Image Readout Losses different storage cell configurations In the vertical frame transfer design, the Readout MTF = cop I -Ne (1 - COS 12irf /f0) j formed by the transfer electrodes in the picture is integrated for one field (16.66 image area (256 actual cells electronically milliseconds). The picture is then read out interlaced 2:1 to achieve 512 sample completely each field. There are therefore VC1 0C2 0C3 positions in the vertical direction). no smearing effects due to lag during PREAMP Resolution is usually defined in terms of panning except for resolution loss due to Television Lines/ Picture Height (tvl/ ph). image motion during exposure as in any 320 STAGE REGISTER

In the horizontal direction, this number is movie film camera. Some other types of 0131

3/4 of the horizontal cell count or 240 solid -state sensors integrate for a full OB2 256 x 320 tvl/ ph due to the 4:3 aspect ratio. In the frame (33.3 milliseconds) and have twice 0133 STORAGE AREA vertical direction, only 486 resolution the panning resolution loss because of the samples are displayed out of the 512 longer exposure time. possible, and the finite line structure in is quoted as reducing the 256 x 320 the display often IMAGE AREA Dark current 0A1 effective useful resolution to ap- (MAY BE INTERLACED 0A2 2:1 VERTICALLY proximately 350 tvl/ ph due to the kell During the operation of a CCD imager, a 0A3 factor even though actual resolution may two -dimensional charge pattern replica be seen to the 486 sample limit with of the light pattern is formed inside the Fig. 1 - Layout of 512x320 element CCD imager. proper pattern phasing. In the horizontal, CCD. Thermally generated charge the video is smoothed into a continuous partially fills the potential wells and waveform and the finite cell (moire) iii'ii 'iia°M.'ïii=t.ssa generates a background signal known as MEASURED BACKSIDE +.Ç=iu effects are not as severe. iii dark current. There are three basic dark - ii - =; iÌ Ì=; Another factor affecting picture current generating mechanisms'. These r

is reduction MTF (sine are illustrated in Fig. 3. The first MEAS s ÉL sharpness the of sources ' CTRODE SIDE ¡liii wave response) due to resolution loss source is diffusion current from the sub- mechanisms in the image formation and strate. This current source can be iIlllIIIIIIikh. IIIi readout processes. These losses are neglected. The second source of dark ...1 . we s. tabulated in Table III for a 1.2 mil cell current is from the depleted layer between ::: interface and the substrate. A dimension. The lens is the first con- the Si02 1 11 tributor to the loss of resolution in image simplified equation for this current densi- L0346-1 #14 Ip1 'THINNED MAGER formation. The lateral diffusion of elec- ty is Jac, = eN,D /2r, where r is the INO ANTIREFLECTION COATING trons in the imager before collection is the effective lifetime and N; is the intrinsic 1 Y. r second mechanism. It is a Sech function carrier concentration. The effective of the optical spatial frequency and field - lifetime is a complex function of the 6 WAVELENGTH IM CRONS) free diffusion length in the bulk. The actual hole and electron lifetimes and the level the centers. If numbers shown are representative of energy of generation Fig. 2 - Spectral response of CCD imager; wavelength is typical geometries. The third loss is the CCD's are processed to remove most of in micrometers. Sin X/ X loss due to the finite cell the generation centers near the center of

sampling process. All of these effects are the bandgap, this component of dark 5102 the -current GENERATION IN also present in silicon vidicons'. The current is less than third dark DEPLETION LAYER

resolution losses in the readout process source. P TYPE SILICON SUBSTRATE are given by the equation in Table III. Ne DEPLETED is the source is genera- is the transfer loss and.fr, frequency The third dark -current SURFACE EE GENERATION of one cell. This resolution loss tion current from surface states at the mechanism is analogous to the resolution silicon S; 02 interface. A simplified equa- GATE DIFFUSION ELECTRODE '.CURRENT loss of a scanning electron beam in a tion for the surface -generation current D GENERATION silicon vidicon. The overall MTF is the density is J, = e N,S/ 2. S is the effective //iG% product of the image formation MTFand surface generation velocity and is the readout MTF. The squarewave proportional to the surface -state density response C 7 r is larger than the sine -wave Nt, near the middle of the bandgap. The response MTF. The actual measured value of Ns., near the center of the band Fig. 3 - Dark current sources in CCD imagers.

www.americanradiohistory.com RESET Table IV TRANSISTOR accumulation. This forms a temporary - Summary of developmental 512x320 CCD imager performance parameters. DRAIN IVR) extension of the channel stop around SAMPLING each charge intergration collecting site 11) Spectral Response Similar to Silicon Vidicon RESET TRANSISTOR GATE (2) Operable in SIT -CCD Mode o-i DRAIN (STD) minimizing charge spread down the 10R) channel. (3) 500 Nanoamp DC Maximum Signal (4) 4 Nanoamp DC Dark Current at 250 C (5) Compatable With Standard 525 Line TV CCD SAMPLING TRANSISTOR (6) Supplies Full Resolution of Color TV (240 TV L /PH) SIGNAL SOURCE (STS) On -chip video processing (7) No Lag - Picture Erased in One Field EXTERNAL FLOATING LOAD RESISTOR DIFFUSION The charge signal in a CCD is VIDEO OUTPUT Notes: manipulated around without touching a SUBSTRATE -1 30, N- Channel Vertical Frame Transfer IVsebl - finite electrode until it is extracted at the 163, 840 1.2 x 1.2 Mil Cells A is Fig. 4 - On -chip video processing. output. floating diffusion used as a 6 MHz Data Rate, 16.66 ms Integration charge detector. The voltage on the 12 MM Image Format floating diffusion is reset to a fixed 500 x 750 Mil Chip Size may differ significantly from the density potential once each clock period by a of surface states nearer the band edges reset transistor. When each charge packet camera tubes for many 525 -line television which affect transfer efficiency. Actual reaches the floating diffusion, the charge applications. It is capable of supplying dark current measurements have yielded changes the voltage on the capacity at the full resolution of color tv (240 tvl/ ph). dark current densities of 5 nA /cm2. This that node. This voltage is sampled by a The features of CCD imagers that are results in a dark current of about 4 nA in sampling transistor operating as a source superior to camera tubes are signal -to- the image sensing area of a 512X320 cell follower. These circuits are shown in Fig. noise ratio, freedom from lag, and device. 4. The very low node capacity of the absence of microphonics. The small size, floating diffusion results in a significant light weight, low -power consumption, The second and third dark- current improvement in signal -to -noise ratio over and precision image characteristics are sources depend on the temperature varia- a silicon vidicon operating at the same additional benefits. Table IV summarizes tion of Ni. This results in the standard light level. In fact, at normal signal levels, the performance parameters for this silicon dark -current variation of a factor no noise is visible in a displayed picture developmental 512 X 320 CCD imager. of 2 for every 9° - 10°C temperature generated by the CCD imager. change around room temperature. References Developmental camera

Blooming I ..Rodger. III. R.L.. "Beam Scanned Silicon Targets for A Camera Tubes," Paper presented at IEEE Intercon, New developmental black- and -white York. NY. March 1973. CCD's exhibit unique problems of charge camera has been fabricated for the 512 X 2. Weimer. P.K., Pike, W.S., Kovac. M.G.. AND Shallcross. F.V., -'The containment when overloaded with light. 320 CCD imager. Fig. 5 shows a picture Design and Operation of Charge- Coupled Image Sensors." Paper presented at IEEE Solid State Circuit is Charge fairly well confined from of this camera. The camera is ap- Conference. Philadelphia. Pa., February 1973. spreading sideways by the channel stop proximately the size of a pack of 3. Kovac. M.F.. Shallcross, F.V., Pike. W.S., and Weimer, diffusions. However, excess charge l'. K., "Design. Fabrication. and Performance of a 128 X 160 can cigarettes and has a C -mount for lens Element Charge Coupled Image Sensor," Paper presented spill up and down the channel. A special interchangeability. Fig. 6 and Fig. 7 show at CCD Applications Conference. San Diego. California. September 1973. mode of operation has been developed to 525 -line monitor pictures made by this 4. l ompsett. M.F.- Amelio, G.F., Bertram. W.S.- Buckles. greatly reduce the spreading of charge camera with a 512 X 320 CCD imager. K.R., McNamara. W.I. Mikkelsen. Jr...1.C.. and Sealer. down the channel under D.A., "Charge-Coupled Imaging Devices: Experimental overload con- Results." IEEE trans. Electron Devices, Vol. ED -IN, ditions. This low- blooming mode is ac- pp.992 -996. November 1971. complished by maaintaining the phase 5. Sequin. CH., "Interlacing in Charge -Coupled Imaging Summary and conclusions Devices," IEEE 'trans. Electron Devices. Vol. ED -20, pp. fingers adjacent to a charge integrating 535 -541, June 1973. phase finger at a voltage that maintains A CCD imager has been developed that 6. Rodgers, 3rd. R.L., et u/, "Silicon Intensifier Target Camera 1 ube." Paper presented at IEEE International Solid State the surface under these fingers in light will offer an attractive alternative to Circuits Conference. Philadelphia, Pa.

Fig. 5 - View of the black- and-white camera for the Fig. 6 - A 525 -line picture made with the 512x320 CCD Fig. 7 -A 525 -line picture made by the 512x320 CCD 512x320 CCD imager. imager. imager.

www.americanradiohistory.com STARTING MATERIAL'' HIGH- RESISTIVITY Power switching n - TYPE SILCON n -TYPE SILICON using solid -state relays DIFFUSE DEED p LAYER ON BOTH SIDES OF T. C. McNulty WAFER

02 Solid -state relays make use of a semiconductor device for control of ac or dc power. Since, in most ac applications, the semiconductor element chosen for power control is the triac, ry GROW SILICON DIOXIDE this paper describes the triac as a power- switching element. Advantages and disadvan- FILM. DEFINE AND D'FF'.ISE p+ REGIONS. tages of the active element over the electromechanical relay are discussed in general terms. Basic parameters, such as surge in -rush capability, transient -voltage ratings, networks, turn -off consideration and the different modes of triac gating are s 02 suppression mammal also discussed. Various circuit designs for on /off control, zero -voltage switching, and line - GROW SILICON DIOXIDE FILM DEFINE AND voltage isolation are described which demonstrate ac power control. DIFFUSE n+ REGIONS.

D+ P+

MGM =MIMI DEFINE AND ETCH POWER SWITCHING using elec- Perhaps, the SSR is to the elec- GATE MOATS AND GRIDS. tromechanical relays is probably as old as tromechanical relay as the transistor was the electrical industry itself. The elec- to the electron tube. That is to say, there J D+ n+ tromechanical relay comes in many forms will not be an immediate widespread (general purpose, telephone type, TO -5, conversion to SSRs, but rather a gradual APPLY AND FIRE HARD reed, mercury wetted, etc.) and has of the new device, particularly GLASS PASSIVATING adoption LAYER. various contact configurations. It is for applications in which increased probably fair to say that most electronic reliability is required and shock or engineers have, at some time, used or mechanical fatigue impose severe designed circuits employing electro- limitations on the electromechanical OPEN CONTACT AREAS relays. During the past few AND METALLIZE. mechanical relay; some movement has already been NICKEL - LEAD -TIN years, the electromechanical device has made in this direction. Probably the SOLCER. LASER- SCRIBE ANC BREAK INTO been challenged by a new breed of relay major limitations to SSR use are cost, PELLETS. which has no moving parts, is capable of line isolation, immunity from line tran- in the fabrication of a handling large amounts of power for sients, and the need for multiple -pole Fig. 1 - The seven basic steps required relatively low input power, and comes in arrangements. However, with solid -state standard, glass -passivated triac. many package and circuit configurations. technology advancing at such a rapid This new solid -state relay (or SSR) uses rate, lower -cost triacs with improved transistors for dc power control or triacs voltage ratings will become increasingly for ac power control. available, and economical methods of line isolation will be developed. With these innovations, the disadvantages of Tom McNulty, Manager, Applications Engineering, Solid SSRs may begin to fade. If so, we may State Division, Somerville, NJ, received the BSEE in 1966 from Newark College of Engineering. Prior to joining begin to see an increase in the use of RCA in 1966, he gained extensive experience in digital SSRs, and the need for a better un- logic- system design at Burroughs Corporation. His first derstanding of triacs as the power ele- assignment with RCA was in Mountaintop as an Applications Engineer with Thyristor /Rectifier ment in ac applications will become a Engineering. In 1967 he transferred to Somerville where must. he is now engaged in the devleopment of circuits for ac power-control. Mr. McNulty hold a U.S. patent and has GLASS PASSIVATION .209 written several paperson the use of SCR's and triacs in TR IAC construction ac power -control and appliance -control applictions. 192 -x.120 Thyristors (silicon controlled rectifiers F.7 and triacs) are semiconductor switches whose bistable state depends upon the 036 .030 030 regenerative feedback associated with a 0 pnpn structure. The SCR is a unidirec- tional device used primarily for dc and ac functions, whereas the triac is a bidirec- 40 AMPERE 8 AMPERE 6 AMPERE IN DESIGN NOW AVAILABLE NOW AVAILABLE tional device used primarily for control of ac power. Fig. 2 - An isometric view of a completed triac and dimensions of three devices now available or in the design stage. The fabrication of a standard, glass -

passivated triac requires the seven basic Reprint RE- 20 -1 -17 (ST -6141) steps illustrated in Fig. 1 and delineated Final manuscript received November 30, 1973.

www.americanradiohistory.com below. rates device surge capability from single Although the transient -voltage problem cycle to multiple cycles. I) The process begins with an n -type, high - Since this rating may seem critical, there are precautions resistivity, silicon wafer; cannot be exceeded repeatedly, care that can be taken to minimize it. The use 2) p layers are diffused deeply into ooth sides; should be exercised in the actual applica- of RC snubbers in parallel with the triac tion to provide a sufficient safety margin 3) Silicon- dioxide diffusion masks are grown, can reduce the rate of imposed transients. and p' regions are defined and diffused into between the published ratings and the This arrangement is most effective for the wafer; actual circuit in-rush currents. fast -rising, short- duration line distur- 4) A second oxide diffusion mask is grown, Another important parameter associated bances. For critical applications, the use and n' regions are defined and diffused into of a voltage -clipping device in addition with a triac is its di/ dt rating, a parameter to the wafer; an RC snubber effectively suppresses 5) A silicon- aioxide etch mask most significant during turn -on. With the is grown and both the rate of rise and magnitude of defined. Grids and gate moats are etched initiation of a gate signal, the active area line-generated transients. into the wafer; closest to the gate region is, essentially, 6) A hard glass- passivated layer is applied in turned on, and, for a few microseconds, Another type of transient particularly the grids and gate moat; the instantaneous power dissipation is a prevalent in the area of inductive loads, 7) Contact areas are opened on the wafer and function of the rate of rise of the on -state and often overlooked, is the nickel -lead -tin solder circuit - metallization is current. This power dissipation may applied. The wafer is then laser- scribed and induced transient. Consider an inductive separated into pellets. cause localized heating and result in load in series with a triac and RC snubber silicon -lattice destruction and triac network which also includes a switch for Fig. 2 contains an isometric view of a degradation. The di /dt ratings are a line -voltage interruption. With the triac completed triac and dimensions of three function of triac geometry and pellet size, in the off state, a leakage current flows devices now available or in the design and ratings of 100 A/ ,us are easily which is a function of the characteristics stage. achieved. In most circuit applications, of the load, the RC snubber network, and stray or actual -load inductance is present, triac leakage. If the switch is momentarily Voltage and temperature and for the condition of di /L, it opened when the triac is qff, then a ratings dt = Epk is easily seen that a few microhenries of voltage transient (E = L di /dt) is inductance are all that are required to generated which can exceed the voltage The effects of voltage and temperature limit circuit di /dt to within the maximum rating of the triac; cause non -gated turn - are important in thyristors because of the rating. When di ratings are exceeded, on and abrupt energy transfer; and may regenerative action of these devices, and /dt it is usually result in damage to the triac. Again, the because they are often required to sup- because of the RC snubber proper selection of RC- network com- port high voltages under high network in parallel with the triac. In such ponents and voltage -clipping device will temperature conditions. The imposed networks, stray inductance is essentially zero, and the magnitude of discharge suppress the circuit -induced transient to a voltages create a field at the junction level compatible with the voltage rating interface, and the increased temperature current is limited by the snubber resistance. The di/ dt in the is of the triac. releases additional surface ions. Should snubber not affected by the inductance added to qwell the field concentrate the additional sur- Commutating dv /dt face charge and allow it to migrate into the di /dt caused by the stray or actual - load inductance; only careful the gate region, non-gated turn -on may selection of The term turn -off time is not associated RC- snubber -network occur. Most manufacturers realize that components will with triacs since triacs are bidirectional, eliminate this second the gate region must be terminated for source of di/ dt and and reverse voltage is nothing more than minimize triac failures. high voltage/ temperature operation, and a forward voltage to one-half of the triac a shunt resistance is built into the triac Transient voltages chip. A new term - critical- rate -of-rise- pellet during fabrication. This shunt of- commutation- voltage - is used with reduces the immunity of the triac to non - It is well known that triacs are susceptible triacs. The term describes the ability of gated turn -on. Additional reliability can to non -gated turn -on and possible the triac to turn off as the current passes be gained by operating the triac under less damage as a result of transient voltages. through zero, or commutates. One must severe voltage/ temperature conditions. Transients are generally caused in a triac remember that the triac is a current - by the switching of inductive loads on dependent device: Current is iniected into In -rush currents adjacent lines or in proximity to the the gate to turn the device on, and current device. If the transient voltage generated must be removed or allowed to pass One of the features that has made exceeds the critical rate -of -rise of the off - through zero for turn -off regardless of the thyristors the work -horses of the power state voltage (dv /dt) then a displacement source -voltage polarity. Commutating semiconductor industry is their ability to current (i = Cdv /dt) is generated which dv/dt is less critical with resistive loads absorb in -rush currents many times in causes non -gated turn -on. Non -gated and most important with inductive loads. excess of their steady -state ratings. This turn -on is not destructive if the energy Consider an inductive load in which the unique feature results from the transfer is within the maximum rating of load current lags the source voltage by a regenerative action of the thyristor, an thé device; however, if the transient phase angle O. As pointed out, triac action which maintains the internal beta voltage does not exceed the off -state commutation occurs at zero current, at a level such that, under in -rush con- dv /dt rating, but does exceed the max- whereas the source voltage has some ditions, the charge density is equally imum voltage rating, then triac breakover magnitude E. As the load current crosses distributed over the entire triac pellet. occurs. Whether triac degradation occurs the zero point, a small reverse current is The equal charge distribution assures the depends on whether the energy transfer is established as a result of the charge in the presentation of a low impedance to the in- within the bulk silicon or the edge n -type region. This charge, plus a dis- rush current. Each manufacturer clearly avalanche. placement current (i= Cdv /dt) resulting

www.americanradiohistory.com is from the reapplied source voltage, can current applied to the load. Actually, ;_ -120-VOLT LINE ..x105 cause the triac to turn on in the absence of there will be leakage- current flow to the t a proper gate signal. A minimum com- load; the amount of current is dependent 2 \,O 10 mutating dv /di at rated current and at a on the applied voltage and triac case i3 6 _- :;; i> S specific operating case temperature temperature. However, because the i 6 _ li-IIIr -- x I 2 \ should be defined in all triac applications; current is very small (less than one

Z 3 the circuit designer can use these milliampere) compared to the load á 0.1 10 I¢ s °lw/..- ` i_T'II a specifications to it be in choose an RC snubber current, can neglected this and the á ; ',Ia2' network that will limit the reapplied following circuits. (In specific dv /di to within ratings. Loss of triac applications in which leakage current ? ; o.016 . Ail .i/.!. 10 control as a result of commutating dv /dt may affect the control unit, it would have i1 a ' ¡ e I does not degrade the characteristics of the to be considered.) VALUES ON CURVES ' I triac. Proper RC snubber network ARE IN V/µs 0.001 / 110 a 6 ° 6 8 selections for worst IC -case conditions of To apply power to the load in Fig. 4, 0.1 , .- load power factor, current, and voltage LOAD CURRENT- A switch SI is closed to provide gate drive la) i2CS-?i?36 are easily made by use of the charts shown to the triac. Bias- resistor R I is of the in Fig. 3. order of 68 to 100 ohms and provides the ::;= 220-VOLT LINE =105 initial gate drive during every half cycle of applied voltage. The power consumption z amen .N Advantages of SSRs 8--=, / O 10a of Rl is very low (1/4 to IA W) because, SN when the triac is in the on state, RI is in i , redoMMINII/ _ 2 AIMM Before the advantages of SSRs are dis- parallel with the on-state voltage of 2-E ``` á .7r 3 U approximately 1.5 V. This method of .- O.I s 10 z cussed, the types available should be 5 6= wl... reviewed. triac triggering, called anode firing, is an _ effective way of triggering because it uses U 2 001!14 Two types of SSR are available: all solid - the source voltage as a source of gate - O.,0` I state and hybrids. The solid -state class current drive. Maximum gate current is nN °í. , employs solid -state devices for both logic available for triac turn -on at peak line VALUES ON CURVES and triac gating. Hybrids generally use a ARE IN V/µs voltage until the device goes to the low - 0.01z'l'/1/ 10 ° F. A 4 E. P

reed relay for triac I gating for ac power impedance state. In this state, the current D.t IO 00 so LOAD CURRENT --A control and combine the electro- in Rl is reduced by the forward voltage mechanical with solid state. In either drop. In effect, bias resistor R 1 is utilized Fig. 3 - (a) Snubber components for 200 -V peak on 120 -V line; (b) class, the triac is used as the solid -state only during the initial turn -on of the triac, Snubber components for 400 -V peak on 220 -V line. element for ac power control. A com- or for approximately two microseconds. parison of SSRs with electromechanical In a typical application, switch SI would relays is given below: be replaced by a relay, and power control would be transferred by means of low - Life: An EMR physically makes and breaks level- current relay contacts. load current, and the relay contacts deteriote with life. SI For control applications which require SS Rs: They have no moving parts, and may be that variable power be delivered to a load, designed to make and break at zero current. Regardless of the design, the triac always an inexpensive RC phase- control circuit breaks at zero current. is best. Fig. 5 shows the basic triac -diac Contact hounce: Inherent with an EMR control circuit with the triac connected in zero for SSRs. series with the load. During the beginning LOAD VOLTAGE

1 RFI: Inherent with EM Rs dependent on of each half cycle, the triac is in the off \N\P SSR design. state; as a result, the entire line voltage is AFL: ( "audio -frequency" interference). Terri- impressed across it. Because the triac is in TRiAC VOLTAGE ble with EM Rs particularly when many with the relays are clacking parallel potentiometer and about. Not noticeable Si 52 with SSRs. capacitor, the voltage across the triac OP F N CLOSED Environment: High humidity, corrosion, and drives the current through the explosive atmospheres usually dictates a potentiometer and charges the capacitor. Fig. 5 - RC phase control. variable power. sealed relay. SSRs may easily he potted. When the capacitor voltage reaches the Shock: The SSR is far superior. breakover voltage of the diac, Veo, the Input logic: EM Rs can be operated from low - capacitor discharges through the triac level logic. SSRs are design dependent, but gate and turns it on. The line voltage is offer complete versatility. then transferred from the triac to the load for the remainder of that half cycle. This sequence is repeated for every half cycle General control circuits of either polarity. If the potentiometer resistance is reduced, the capacitor A simple triac control circuit, an on /off charges more rapidly, the Veo of the diac circuit, is shown in Fig. 4. With switch SI is reached earlier in the cycle, and the open, the triac is off and essentially zero power applied to the load is increased. If Fig. 4 - ON /OFF control. non -synchronized.

www.americanradiohistory.com the potentiometer resistance is increased, proportional control differs from the triggering occurs later and load power is on/ off control in that it allows a specified reduced. The main disadvantage of this percentage of power (duty cycle) to be circuit is that it produces RFI. supplied to the load with a finite off time that, in turn, allows the heating element Although the basic light -control circuit to "catch up" as a result of thermal lag. In operates with the component arrange- effect, this scheme provides "anticipator ment shown in Fig. 5, additional corn - control." Again, the key to circuit opera- ponents and sections are usually added to tion is in the state of the differential NTC NEGATIVE TENOERTURE COEFFICIENT reduce hysteresis - effects, extend the amplifier. Fig. 6 -b ock diagram of the integrated- circuit zero -voltage switch, effective range of power control, and CA3059. suppress radiofrequency interference. AC -line isolation

Temperature -control circuits The design engineer often must provide dc-to-ac isolation. Complete isolation A zero -voltage switch, Fig. 6, syn- can be achieved by reed relays, pulse chronized for line -pulse generation, in transformers, and light- activated devices. combination with a triac, is particularly Selection of any one of these three well suited for temperature -control approaches depends on the dc logic applications. The zero- voltage- design and component economics. Fig. switch /triac circuit may be used with an 8(a) shows a reed relay and transistor on /off-type control or as a proportional drive circuit which is effective in triac control depending on the degree of gating, although it does have moving regulation required. A simple, inex- parts. Fig. 8(b) uses a pulse transformer pensive, on /off temperature controller is for isolation, and requires a form of clock shown in Fig. 7; a review of the functional pulse that can be transferred to the triac Fig. 7 - CA3059 ON /OFF temperature controller. block diagram of the zero -voltage- switch, gate. In some applications, clock pulses Fig. 6, will help in understanding the may already be available; therefore, the REED SWITCH circuit. For every zero -voltage crossing, a pulse- transformer approach is zero crossing pulse is generated and economical. This approach requires more directed to the triac gating circuit. If there components than that of Fig. 8(a), but it is a demand for heat, the differential has no moving parts. The last approach, amplifier is in the open state, the triac and, at present, probably the most ex- gating circuit is open, and the triac is pensive one, uses a light- activated device, INPUT turned on at every zero- voltage crossing. such as the GaAs infrared emitter, to When the demand for heat is satisfied, the initiate triac gating. The light- activated differential amplifier is in the closed state; device is coupled to a photosensitive this inhibits the triac gating circuit and transistor which, when turned on, removes any further gate drive to the provides inhibit logic for additional in- triac. Therefore, the key to the operation tegrated circuits or, as in Fig. 8(c), for a of this circuit is in the state of the zero -voltage- switch application. differential amplifier. One side of the

CLOCK PULSE differential amplifier is biased to a 5 kHz reference voltage VR, and the other side is Conclusion

SI biased to a voltage Vs which is dependent on a variable potentiometer setting and The designer who thoroughly un- sensing resistor. As a result, whenever the derstands the characteristics and bias voltage Vs exceeds the reference limitations, but most of all the advan- voltage VR, the gating circuit is open and tages, of triacs will have at his disposal a the triac is turned on for each zero - device that he can use to design power voltage crossing. The characteristics of an controllers that operate satisfactorily not on / off controller are well known; i.e., only in normal applications, but also in there are significant thermal overshoots severe physical and electrical and undershoots which result in a environments. The triac has already proven to be a true power- semiconductor I20 VAC differential temperature above and below the reference temperature. The device, and is widely used in both corn - magnitude of the differential temperature mercial and industrial applications; is dependent on the mass of the heater restrictions on triac use in military and the time constant of the sensing applications, particularly in 400 -Hz FROM _OGIC power systems, are gradually being lifted. SOURCE element. It is inevitable, then, that the triac will For precise temperature control, the evolve as the basic building block for ac technique of proportional control with power control in power -controller Fig. 8 - (a) Isolation with reed relay; (b) isolation with pulse transformer; (c) isolation with light- activated devices. synchronous switching is introduced. The systems.

www.americanradiohistory.com Engineering and Research Notes

Adhesive bonding of semiconductors to sub- cure permits sufficient time tor any flow of epoxy that might he required to fill in voids and thin regions created when the wafer is strates pressed into the layer. However, the curing rate is fast enough to produce sufficient bond strength for further processing after an overnight room temperature cure. After the overnight cure. the wafer Robert D. Larrabee and substrate are sufficiently bonded to permit removal of excess Advanced Technology Laboratories adhesive. After cleaning, the device can be put in an oven to quickly Somerville, NJ effect a complete cure.

This process has routinely yielded 2- to 3 -gm epoxy layers that are bubble -free over the entire (Hg-Cd)Te wafer surface area of approximately I cm'. They have bond strengths exceeding the strength of the (Hg -Cd) Te itself. Since this process is not dependent on any particular property of either sapphire or (Hg- Cd)Te. it should be applicable whenever two flat, polished surfaces are to be bonded When fabricating infrared imaging arrays, we were faced with the together by a thin adhesive layer. problem of attaching a delicate (Hg -Cd)Te semiconductor wafer to a sapphire substrate with a thin adhesive layer that had to be both uniform in thickness and bubble -free. Our solution to this problem may to prove Material References be a useful technique for a variety of unrelated applications.

The most effective procedure involved coating the polished sapphire I. Eccohond 451.V (clear). Emerson and Caning. Inc.. Canton. Mass. substrate with a thin (several gm) layer of epoxy resin, gently pressing 2. Catalyst 15LV (clear). Emerson and Cureing. Inc., Canton. Mass. the polished (Hg -Cd) Te wafer into the epoxy layer under vacuum, and 3. Aiosol Brilliant Yellow. OAF Corp.. 141) West 51st Street, New York. New York (Then) is holding it together at room temperature until the epoxy was sufficiently nothing unique about this dye other than its intense fluorescence when dissolved in the epoxy resin. A number of other dyes will serve the purpose equally well..) cured to permit removal from the vacuum apparatus without disrupting the bond. The epoxy layer was produced by mixing Eccobond 45LV Resin' and its hardener, Catalyst 15LV,' with acetone in such a Reprint RE- 20 -t -25] Final manuscript received October 19, 1973. concentration that when a few drops of the mixture were applied to the sapphire substrate, the desired thickness of adhesive was left after the acetone diluent evaporated.

It was found that the sapphire substrate had to be very carefully cleaned and had to have a lyophilic reaction to acetone. If these conditions weren't effected, the resin mixture would ball -up before all of the Millivolt source for temperature programming acetone was evaporated, and a series of epoxy "islands" would result of laboratory furnaces instead of a thin uniform layer. A rough (i.e., forepump) vacuum was necessary to eliminate any trapped air pockets that would otherwise inhibit the epoxy flow into, or out of, any irregularities in the substrate. R. Fehlmann I A.E. Widmer layer, or semiconductor wafer. adhesive Laboratories RCA Ltd. Zurich, Switzerland Since the thin adhesive layer is so very transparent, it proved difficult to optically judge the uniformity of its thickness. Therefore, a fluorescent dye` was incorporated into the original diluted epoxy mixture to allow for inspection of the thin resin layer under ultraviolet light (366 -nm Hg line). Uniformity was subjectively judged by the intensity and uniformity of the fluorescence emission from the epoxy layer. Fehlman Widmer

The recipe for the actual mixture used is: A simple, inexpensive programmable millivolt source can be used for linear rate- controlled raising or lowering of a laboratory furnace Acetone 200 cc temperature. Two versions of such a source have been designed and are Epoxy resin' 5 cc described in this note. Hardener' 5 cc Fluorescent dye' 25 mg The output of either programmer is set for typical applications such as the controlled growth of Ill -V compounds (GaAs, GaP, etc.) by liquid - phase epitaxy for which Pt -Pt 10 %o Rh sensing thermocouples are used. This mixture has a shelf life of several months when stored in a brown The ranges are as follows: bottle at room temperature. After the diluent evaporates, the undiluted 1)5 to 50 gV /min. or 0.45 to 4.5 °C /min. (950 to 1000 °C) epoxy layer requires about 24 hours to cure at room temperature (less time, of course. at elevated temperatures). This slow room- temperature 2)25 to 250 gV /min. or 2.2 to 22 °C /min. (950 to 1000 °C)

www.americanradiohistory.com 15V111- NTEGRATOR ISOLATION COMPARATOR 15V STAGE 415 V P2 15 V

MOUNTED ON TEFLON BOARD & COVERED WITH PLEXI ROG

R6 495 I 2 -20%

PI Como. 500 mm IOOmv /MIN

SLOPE XLI

L 4 10K 1%

OV ImAma, 55

25 . 0V pp1Y máUT25mv mo. 250YV/wN. 25 1%

a MOUNTED ON TEFLON BOARD OUTPUT FOR 25mV B COVERED WITH PLEXI BOX mo. 250 YV /MIN

Fig. 2 Circuit diagram for a programmable millivolt - Fig. 1 - Circuit diagram for a programmable millivolt source for rate - - source for rate -controlled heat -up, hold, and cool controlled cool -down processes. (OA1 - Zeltex ZA 801 M1; R3 - down processes. (OA1 - Analog Devices 41 K; 0A2 - Texas Instrum, SN 72741; 0A3 - Texas Instrum. Glass encapsulated resistor 101° ohms.) SN 72810; R3 - Glass encapsulated resistor 1010 ohms.)

The simpler version is for application where only a rate -controlled temperature set at the temperature controller can be chosen with the lowering of the temperature is required, while the more versatile version potentiometer P5. The hold voltage from this potentiometer and the allows a rate-controlled raising and lowering of the temperature, and in ramp voltage from P3 are compared at the comparator OA -3. The addition has a preset hold control that can be set to maintain constant comparator switches its output voltage from 0 to + 3.2V when the ramp temperature. voltage is equal to the preset hold voltage. This actuates a switching circuit with the transistor TI and the relay RL2. An isolation stage (OA- As shown in Fig. I, the simpler version has only one FET- operational 2) is required between the ramp generator and the comparator to amplifier which can be used where only rate -controlled cool -down of the prevent loading of the ramp generator by the switching current of the temperature is required. A more versatile version for rate -controlled comparator. heat -up, hold, and cool -down with three operational amplifiers is shown in Fig. 2. Both circuits can be built with readily available low- priced Other circuit components are: RI,' ,2, reed relays with an isolation operational amplifiers and a few passive components. The material cost resistance of 101` ohms; R3, glass encapsulated resistor(e.g. Welwyn or for the programmer shown in Fig. I is about $ 50 and for a more versatile Victoreen): R4,5,7, metal film resistors; P2,3,4, trimpots, 20 turn. The version (Fig. 2) about $ 100. programmers require a regulated ± 15 V power supply with 100 mA

output for circuit 1 and 250 mA for circuit 2. Both circuits contain a ramp generator for the rate controlling process. The output voltage versus time for such an integrator with a constant input voltage is Special attention must be given to the wiring of the integrator and its associated components. Any isolation losses will affect the long -term stability. All these components are mounted separately on a Teflon I t Vlt board and are encased in a Plexiglas box. V2 = -- f V,uldt=-- RC o RC The front panel controls for the simpler version include the ramp voltage potentiometer PI, the range switch S2, the percentage span meter Il, High accuracy for long ramp periods requires that the capacitor be a and switches S3 for normal and fast run and SI for run -down and reset. low- leakage type, that the have a operational amplifier high input The slope potentiometer PI on these programmers can in addition be impedance, and that the leakage of the passive currents other coupled to a synchronous motor which allows the ramp rate to increase components and switches of this circuit be less than IO pA. We used a or decrease linearly. For circuit 2, the front panel controls are the same high -quality metallized polycarbonate capacitor for Cl and an FET as above except that S I is for run -up and run -down and that the hold operational amplifier for OA -I. potentiometer P5 and the indicating lights LED1,2,3 are included.

The ramp rates are set on the potentiometer PI. Both programmers have the same ranges: a low one from about 5 to 50 µV / min and a high one The linearity of the ramp voltage versus time over the whole voltage from about 25 to 250 µV / min. For a Ps-Pt IO % Rh thermocouple span for both programmers shows no deviations or ripples and is within between 950 and 1000 °C, this corresponds to heating rates of about 0.45 the linearity of the recorder used for monitoring. The same was found if

to 4.5 °C/ min. and 2.2 to 22 °C/ min., respectively. The maximum ramp the temperature of a furnace versus time was recorded. For . the rate of the integrator itself is given by the voltage from the potentiometer programmer with the hold feature (Fig. 2) the most critical requirement PI and by the values of the resistor R3 and the capacitor Cl. The is the stability of the hold voltage with time. Any drift in this voltage will maximum ramp rate set with PI corresponds to 100 mV /min. at the result in a drift of the furnace temperature. A drift measurement was integrator output. This rate is reduced by the ratio of the voltage divider made over 75 hours with a hold setting of 50% of the voltage span. Only (R4, R5), which leads to a maximum ramp rate at the output terminals a small, linearly increasing, positive drift voltage of 3 MV /hour was of 250 oV /min. (100 %) and a minimum value for 2% of 5 /IV/ min. found which can be neglected for most practical applications. Small corrections of RI and R2 may be necessary to account for the alloweu deviations of the resistor value of R3 and the capacitance of Cl. We have used these controllers in our laboratory for programming the temperature of liquid phase epitaxial growth of GaP- (GaA!)P The maximum output voltage, or span, is given by the voltage across the heterostructures. We found them very simple to build and to operate. resistor R5 and is set for 25 mV. Other span voltages can be set by These programmers can be easily added to any existing furnace changing the value of R5, keeping in mind that this will also affect tilt controller circuitry since they have to be only connected in series with ramp rates. the thermocouple sensing element.

The additional circuit sections of Fig. 2 consist of an isolation stage, a comparator for setting the hold temperature, and a relay switching circuit. A hold temperature below the maximum processing Reprint RE -20 -1 -251 Final manuscript received November 20, 1974.

www.americanradiohistory.com PRODUCT INTEGRITY through design - LIQUID I Ir , Effect of electron cooling on the A.C. Spear (GCASD,Mass) 12th Annual Spr- lifetime of - R.S. Crandall (Labs,Pr) Physics ing Reliability Seminar; IEEE Boston Section; Letters. Vol. 46A, No. 6, pp. 385-386; 1/28/74 tl 4/25/74 _t;(Ijl1. PHOTOCHROMIC I.iNhO,, Hologram RELIABILITY PREDICTION modelling using storage in - D.L. Staebler, W. Phillips multivariate data analysis methods - F.R. (Labs,Pr) Appl. Phys. Letters, Vol. 24, No. 6; Freiman (ATL,Cam) IEEE Boston Section 3/15/74; pp. 268 -270 Spring Seminar; Boston. Mass.: 4/25/74 SEMICONDUCTORS, Avalanche multiplication in - J. Conradi (RCA Ltd,Mont) Univ. of 180 Management and Business Sherbrokke: 4/4/74 Operations SILVER AND GOLD FILMS, Optical properties of - R.W. Cohen, G.D. Cody, M.D. organization, scheduling, marketing, Courts, B. Abeles ( Labs,Pr) Physical Review personnel. B. Vol. 8, No. 8; pp. 3689 -3701: 10/15/73 t f Recent Pen RCA SrC12, Reduced 4f -5d electrostatic interac- BLANKET ORDERS: A solution to purchas- tion of I M in - R.C. Alig, R.C. Duncan, B.J. technical papers ing problems - J.A. Lyons, Jr. (EC,Har) 65th Mokross (Labs,Pr) J. of Chem. Phys., Vol. 59, Podium and presentations Annual SLA Conf., Toronto, Canada; 6/9- No. 11, 12/1/73 pp. 5837 -5841 and 13/74 STOICHIOMETRIC IRON SULFIDE SINGLE Activities, The role of Both published papers and verbal presentations are indexed. To obtain a published paper DESIGN -TO -COST CRYSTALS near the alpha transition F.R. Freiman borrow the journal in which it appears from your library or write or call the author for a parametric cost estimating in - temperature, Magnetic properties of - T. reprint. For information on unpublished verbal presentations write or call the author. (The ( ATL,Cam) AllE Seminar Design -to -a -cost, Takahashi ( Labs,Pr) Sol. St. Comm., Vol. 13. author's RCA Division appears parenthetically after his name in the subject -index entry.) Washington, D.C.; 3/27 -29/74 1973; pp. 1335 -1337 For additional assistance in locating RCA technical literature contact RCA Technical Communications, Bldg. 204 -2, Cherry Hill, N.J. (Ext. PC- 4256). DESIGN -TO -COST contracting guidelines, SULFO -SPINELS containing iron -group Summary of American Preparedness transition metals, Some new - S. Harada This index is prepared from listings provided bimonthly by RCA Division Technical Association Report on - C.D. Fisher ( Labs,Pr) Mat. Res. Bull., Vol. 8; 1973; 1361- Publications Administrators and Ed,torial Representatives -who should be contacted (GCASD,Cam) Mtg. Proc., ADPA, p. 10 -1 thru concerning 1370 errors or omissions (see inside back cover) 10 -15; 10/15 -16/73 Subject index categories are based upon standard announcement categories used by PROJECT MANAGEMENT - planning, SURFACE POLARITONS at metal surfaces, Technical Systems. Bldg. 204 -2, N.J. Information Cherry Hill, scheduling, and control - M.W. Buckley Dispersion of - R.C. Alig ( Labs,Pr) Solid ( MSRD,Mrstn) Canadian Management Cen- State Comm.. Vol. 13: 1973 tre of the American Management Associations, Toronto, Canada; 4/22 -25/74

PROJECT planning, 130 Mathematics MANAGEMENT - SURFACE POLARITONS at metal surfaces, Subject Index scheduling and control - M.W. Buckley Dispersion of R.C. Alig (Labs.Pr) Solid basic and applied mathematical - ( MSRD,Mrstn) AMA Management Center. State Comm.. Vol. 1, 1973 methods. Chicago, III.; 4/29 -5/2/74 Titles of papers are permuted where SULFO -SPINELS containing iron -group necessary to bring significant keywords(s) to transition metals, Some new - S. Harada MULTIVARIATE STATISTICAL ANALYSIS, the left for easier scanning. Author's' division ( Labs.Pr) Mat. Res. Bull., Vol. 8; 1973; 1361- Use of computerized techniques in B.B. appears parenthetically after his name. - SERIES 200 1370 Bocciarelli ( ATL,Cam) IEEE Coll. RCA MATERIALS, DEVICES, & Burlington. Mass.; 2/21/74 VAPOR -DEPOSITED \h :Sn alloys, Prepara- COMPONENTS tion and high -field superconducting RYTOV'S method and large fluctuations, SERIES 100 properties of - R.E. Enstrom, J.R. Appert Comments on D.A. deWolf ( Labs,Pr) J. of - ( Labs.Pr) J. Appt. Phys.. Vol. 45, No. 1, 1/74; BASIC THEORY & Acoust. So. Am. No 53; 1973; pp. 1109 pp. 421-427 METHODOLOGY 205 Materials (Electronic) VAPOR -GROWN In,(ia ,As Epitaxial Films 160 Laboratory Techniques and preparation and properties of con- on GaAs and In.(ia, ,P substrate, Optical Equipment ductors, semi -conductors, dielec- properties of - R.E. Enstrom, P.J. Zanzucchi, trics, magnetic, electro- optical, J.R. Appert ( Labs.Pr) J. Appt. Phys., Vol. 45, No. 1, 1/74 105 Chemistry experimental methods and equip- recording, and electromagnetic organic, inorganic, and physical. ment, lab facilities, testing, data materials. YTTRIUM IRON GARNET, Light scattering measurement, spectroscopy, elec- from thermal acoustic magnons J.R. tron microscopy, dosimeters. - Sandercock, W. Wettling ( Labs,Pr) Sol. St. CHEMICAL POLISHING of sapphire and Sil , On the existence of bielectrons in - W. Comm.. Vol. 13, 1973 spinet - P.H. Robinson, R.O. Wance Czaja, G. Harbeke, L. Krausbauer, E. Meier, A C B.J. Curtis, H. Brunner ( Labs,Pr) Sol. St. ( Labs.Pr) RCA Review. Vol. 34, No.4, pp.616- FURNACE, simple 2100° 600 -Watt tube - Comm., Vol. 13, pp. 1445 -1450; 1973 Zn- IMPLANTED GaN, Photoluminescence of 12/73 G.W. Webb, R.E. Miller ( Labs,Pr) Rev. Sci. 622; J.I. Pankove. J.A. Hutchby ( Labs,Pr) Instrum.. No. 10, Vol. 44, 10/73; pp. 1542 -1543 - C'dCrSe, Composition of - R. Okamoto, K. Applied Physics Letters, Vol. 24, No. 6, 3/15/74 SPECTROMETRY below 10 keV, Ion scatter- Ametani, T. Oka (Labs,Pr) Japan J. Appt. Phys.. Voll. 13, p. 187 -188: 1/74 ing - R.E. Honig, W.L. Harrington ( Labs.Pr) 125 Physics Thin Solid Films. 1973; pp. 43 -56 DILUTE COSPUTTERED MULTICOMPO- 210 Circuit Devices and electromagnetic field theory, quan- NENT FILMS, Calculation of composition of tum mechanics, basic particles, Microcircuits 175 Reliability, Quality Control J.J. Hanak, B.F.T. Bolker ( Labs,Pr) J. of plasmas, solid state, optics, thermo- - and Standardization Appt. Phys.. Vol. 4. No. 11, 11/73 pp. 5142- electron tubes and solid -state dynamics, solid mechanics, fluid 5147 mechanics, acoustics. devices (active and passive); in- value analysis, reliability analysis, tegrated, array and hybrid micro- standards for design and production. ELECTRONIC MATERIALS, Chemical vapor circuits, field- effect devices, and modular ELECTROGYRATORY EFFECTS, Magnitude deposition of - J.J. Tietjen (Labs,Pr) Ann. resistors capacitors, and printed circuits, circuit inter- of A. Miller (Labs,Pr) Phys. Rev. B 15, Vol. 8, Rev. of Mat. Sci., Vol. 3, 1973; pp. 317 -326 - INTERCONNECTIONS - B.R. Schwartz connection, waveguides and No. 12, pp. 5902 -5908: 12/73 ( MSRD,Mrstn) Electromechanical Design EPITAXIAL InAs, Carrier lifetime in - V.L. transmission lines. IONIC Dalai, W.A.I- licinbothem, H. Kressell ( Labs,Pr) CHARGE at room temperature MANUFACTURING, Electronic equipment through the interface of - Appt. Phys. Lett., Vol. 24, No. 4, 2/15.74 air with Sit:, Injec- D.H. Mercer (MSRD,Mrstn) Proc. of the Dis- AVALANCHE DIODES, Temperature effects tion and removal of M.H. Woods, - R. crete Manufacturing Industries Workshop, in silicon - J. Conradi (RCA Ltd,Mont) Solid Williams ( Labs,Pr) J. Appt. Phys., Vol. 44, No. GaAs:Cs -O surfaces, Calculated energy dis- Report No. 3, 2/74 State Electronics, Vol. 17, 1974; pp. 99 -106 12.; 12/73 pp. 5506 -5510 tributions of electrons emitted from negative electron affinity J.S. Escher, H. Schade RESIDUE FLUXES in historical perspective, - CHARGE -COUPLED DEVICES, Experimen- RAMAN SCATTERING from a nematic liquid ( Labs,Pr) J. Appl. Phys.,Vol. 44, No. 12, 12/73; water soluble O.D. Black (ATL,Cam) tal measurements of noise in - J.E. Carnes. crystal: orientation studies - E.B. Priestly - pp. 5309 -5313 NEPCON West '74, Anaheim, Calif.; 2/29/74; W.F. Kosonocky, P.A. Levine ( Labs,Pr) RCA ( Labs.Pr) Phys. Rev. Lett., Vol. 31, No. 26, pp. Proc. Review. Vol. 34. No. 4, 12/73' pp. 553 -565 1552 -1556; 12/24/73 LANTHANIDES: some lanthanide tungstate- SEMITRANSPARENT MASKS, Fabrication of chlorides, Observation of the crystal COLOR PICTURE TUBES, Three- filter TURBULENT ARCS, Voltage gradients in chemistry of P.N. Yocom, R.T. Smith - - N. Feldstein, J.A. Weiner ( Labs,Pr) J. - colorimetry of - G.M. Eheman, Jr. (EC,Har) I.P. Shkarofsky (RCA Ltd,Mont) Canadian J. Electrochem. Soc., Vol. 120, No. 12,; 12/73 pp. ( Labs,Pr) Materials Research Bulletin. Vol. 8, Electrochemical Soc., San Francisco, Calif.; of Physics, Vol. 52, No. 1, 1974: pp. 68-79 1654 -1657 No. 11; 11/73; pp. 1287 -1294 5/12/74

www.americanradiohistory.com GaAs FET for high power amplifiers at feeds and couplers, phased arrays; Optics in mapping; Rochester, N.Y.; 3/28/74 JET ENGINE ACCESSORIES, Automatic test microwave frequencies - L.S. Napoli, J.J. randomes and antenna structures; system - G.E. Kelly, Jr. ( GCASD,Mass) Hughes, W.F. Reichert, S. Jolly (Labs,Pr) RCA electromagnetic wave propagation, Automatic Testing Conf., - International 74, Review, Vol. 34, No. 4, 12/73; pp. 608 -615 scatter, effects of noise. 250 Recording Components and Brighton, England; 11/5/74 Equipment HETEROJUNCTION DIODES for optical communications, High radiance, high speed ANTENNA installations and circularly disk, drum, tape, film, holographic 340 Communications AI,Ga, -,As M. Ettenberg, H.F. Lockwood. - polarized TV, Multiple M.S. Siukola for im- J.P. Wittke, H. Kressel (Labs,Pr) IEDM Tech. - and other assemblies audio, ( CCSD,Cam) Soc. of Broadcast Engineers age, and Digest, 1973 data systems. industrial, military, commercial Mtg., Phila. Pa.: 2/25/74 systems, telephony, telegraphy and telemetry, (excludes: television, and I_iNbO, transducer for microwave acoustic CIRCULAR POLARIZATION for Im and tv radio). (bulks) delay lines, Fabrication of sub -micron LASER RECORDING on black and white film, broadcast broadcasting, Some aspects of O.Ben -Dov S.L. Corsover (ATL,Cam) -H.C. Huang, J.D. Knox, Z. Turski, R. Wargo, - Multispectral - ( CCSD,Cam) Soc. of Broadcast Engineers, European Electro Conf., Mon - J.J. Hanak (Labs,Pr) Appl. Phys. Lett., Vol. 2, Second -Optic New York; 2/14/74 treau. 4/5/74 MICROWAVE TRANSMISSION study, No. 3, 2/1/74 Switzerland; Digital /analog - K. Feher (RCA Ltd,Mont) CIRCULAR POLARIZATION in fm and tv PHOTOMULTIPLIER TUBES in nuclear QUADRUPLEX VIDEO RECORDING, Recent Doctorate Thesis presented at the Univ. of broadcasting, Advantages of M. Siukola in K. Sadashige medicine, The role of - D.E. Persyk, W.D. - developments - Sherbrooke; 3/74 ( CCSD,Cam) Soc: of Broadcast Engineers, ( 115th Conf., Los Lindley (EC,Har) Soc. of Nuclear Medicine CCSD,Cam) SMPTE Chicago; 2/20/74 Angeles, Calif.; 4/25/74 12th Intl Annual Mtg., Munich, Germany; 9/11 -14/74 REFLECTIONS, Another look at - W. Max- RE- RECORDING, A high speed interlock 345 Television and Broadcast system for L.A...Briel, R.S. Dickinson SCHOTTKY BARRIERS. Granular metal - well (AED,Pr) QST, Vol 57, No. 4 (April 1973) - (CCSD,Cam) SMPTE 115th Cont., Los semiconductor C.R. Wronski, B. Abeles, and in subsequent issues in several in- television and radio broadcasting, - Angeles, Calif.; 4/23/74 R.E. Daniel, Y. Arie (Labs,Pr) J. Appt. Phys., stallments. receivers, transmitters, and systems, Vol. 45, No. 1, 1/74 television cameras, recorders, studio VIDEO TAPE PERFORMANCE, Biased tape equipment. SUPERCONDUCTING TUNNEL guidance for improved - J.D. Bick JUNCTIONS, Current- voltage ( CCSD,Cam) SMPTE 115th Conf., Los 240 Lasers, Electro- Optical and Angeles, Calif.; 4/22/74 characteristics of - W.C. Stewart (Labs,Pr) J. Optical Devices HOLOTAPE: A low -cost prerecorded televi- Appl. Phys., Vol. 45, No. 1, 1/74 sion system using holographic storage - W.J. Hanna, R.E. Flory, M. Lurie, R.J. Ryan design and characteristics of lasers, TRAPATT CIRCUITS, Lumped -element ( Labs,Pr) Soc. of Motion Pic. & Telev. high -power - A.S. Clorfeine, H.J. Prager, components used with lasers, 280 Thermal Components and Engineers Vol. 82, No. 11, 11/73; pp. 905 -915 R.D. Hughes (Labs,Pr) RCA Review, Vol. 34, electro- optical systems, lenses, etc. Equipment No. 4, 12/73 (excludes: masers). TV TRANSMITTING SYSTEMS for unattended operation T.M. Gluyas heating and cooling components - TWO ( CCSD,Cam) 28th NAB engineering conf., -PHASE CHARGE- COUPLED and equipment, thermal measure- and ELECTROLUMINESCENT JUNCTIONS on Houston. Texas: 3/17 -20/74 DEVICES with overlapping polysilicon ment devices, heat sinks. aluminum gates, Design and performance of GaAs, Vapor -Grown in In, ,Ga,P - C.J. Nuese. A.G. Sigai, J.J. Gannon, T. VIDEO RECORDING, A of - W.F. Kosonocky, J.E. Carnes ( Labs.Pr) comparison IEDM Tech. Digest, 1973;pp. 123 -125 Zaeromowdki ( Labs,Pr) J. of Electr. Mat., Vol. AIR CONDITIONING using solar energy, quadruples and helical scan - J.L. Grever 3, No. 1, 2/74 Regenerative gas cycle - B. Shelpuk ( CCSD,Cam) SECA Mtg., in Galt House, (ATL,Cam) National Science Foundation Louisville, Ky; 5/14/74 GaP:N, The effect of hydrostatic pressure on Solar Cooling Workshop; 1/31/74, the photoconductivity and Washington, D.C.; Proposal to National 215 Circuit and Network electroluminescence of - H. Kressel, M.I. Science Foundation Designs Wolfe, T. Halpern, P.N. Raccah ( Labs,Pr) 360 Computer Equipment Applied Physics Letters, Vol. 24, No. 6; 3/15/74 analog and digital functions in elec- processors. memdries, and tronic equipment; amplifiers, filters, peripherals. In,Ga, - , As for 1.06 micrometre emission, modulators, microwave circuits, A -D CW laser diodes and high -power arrays of SERIES 300 converters, encoders, oscillators, - C.J. Nuese, M. Ettenberg, R.E. Enstrom, H. SYSTEMS, EQUIPMENT, switches, masers, logic networks, MNOS MEMORY development J. Richards Kressel ( Labs,Pr) Appl. Phys. Lett., Vol. 2, No. - timing and control functions, fluidic & APPLICATIONS 5. 3/1/74 (ATL.Cam) Hardened Guidance & Weapon circuits. Delivery Technology Mtg., Culver City, Calif.; In,Ga, -, As pn junctions, Room -temperature 3/26 -27/74 ACOUSTIC DELAY LINES utilizing I.iNbO,, laser operation of - C.J. Nuese, R.E. Bulk H.C. Huang, J.D. Knox, Z. Turski Enstrom, M. Ettenberg (Labs,Pr) Appl. Phys. READ -WRITE HOLOGRAPHIC MEMORIES, - Heterodyne readout for R.S. Mezrich, W.C. ( Labs,Pr) 1973 IEDM Tech. Dig., 1973; pp. Lett., Vol. 24, No. 2, 1/15/74 312 Land and Water Transporta- - ( Labs,Pr) Applied Optics. Vol. 12, No. 263-265 tion Stewart INJECTION LASERS: V. Strong polarization, 11, 11/73 pp.2677 -2682 D/A Subsystem Development, Radiation Experimental properties of H.S. Sommers, - navigation systems, automotive elec- hardened - L. Dillon (ATL,Cam) Hardened Jr. ( Labs,Pr) J. Appt. Phys., Vol. 45, No. 1, 1/74 Guidance & Weapon Delivery Technology tronics, etc. Mtg.; Culver City, Calif. 3/26-28/74 LASER DIODES, of In,Ga, ,As for 1.06 365 Computer Programming micrometre emission, Efficient - C.J. Nuese, and Applications DIGITAL INTERPOLATION FILTERS for in- M. Ettenberg, R.E. Enstrom, H.Kressel COLLISION AVOIDANCE, A new kind of radar for J. Shefer, R.J. Klensch, HC. creasing the sampling rate, Parallel ( Labs,Pr) 1973 IEDM Tech. Digest: 1973 - languages, software systems, and realizations of H. Urkowitz (MSRD,Mrstn) Johnson, G.S. Kaplan (Labs,Pr) Soc. of Auto - general applications (excluding: 1974 Intl Symp. on Circuits and Systems OPTICAL PRECURSOR detection, Laser Eng., Automatic Engineering Congress, - specific programs for scientific use). Theory, San Francisco, Calif; 4/22/74 W. Shubert, L. O'Hara, E. Lea. L. Hansen Detroit, Mic.. 2/24 -3/1: 1974 (ATL,Cam) DoD laser conf., Colorado FREQUENCY DIVIDERS, Microwave - L.C. Springs, Col.; 3/28/74; Proc. VEHICLE -IDENTIFICATION system, A microwave automatic R.J. Klensch, J. CASPER (Comprehensive APL system for the Uphadhyayula, S.Y. Narayan ( Labs,Pr) RCA - preparation and execution of reports) W.A. Review, Vol. 34, No. 4; 12/73; pp. 595 -607 RELIEF PHASE HOLOGRAMS recorded in Rosen, H. Staras ( Labs,Pr) RCA Review, Vol. - Rux, E. Baelen (EC,Har) APL 6 Intl User's photoresists, Characteristics of R.A. Bar - 34, No. 4, 12/73; pp. 566 -579 - Conf.. Anaheim. Calif.: 5/14 -17/74 PHASE -LOCKED LOOPS, Burst syn- tolini ( Labs.Pr) Appl. Opt. Vol. 13, No. 1; 1/74 chronization of L. Schiff ( Labs,Pr) IEEE, - DESIGN AUTOMATION: Make it work in a Communications Tech. Vol. COM -21 No. 10; SCATTERING OF LIGHT by phonons in drafting environment A.T. Farrell 10/73 absorbing materials R.K. Wehner (Labs,Pr) 315 Military Weapons and - - (GCASD,Mass) 1974 American Institute for Opt. Comm., Vol. 6, No. 2; 10/72 Logistics SILICON IMPATT OSCILLATORS, FM noise Design & Drafting Conf.. Chicago 4/3/74 measurements on p-type and n -type G.A. - missiles, command and control. SOFTWARE COST performance, Automatic Swartz, Y.S. Chiang, C.P. Wen, A. Young 245 Displays test equipment - H.L. Fischer ( Labs,Pr) Electronics Letters, Vol. 9, No. 25; (GCASD,Mass) Aeronautical Systems 12/13/73 AEGIS weapon system - F.G. Adams Software Workshop, Dayton, Ohio: 4/3/74 equipment for the display of graphic, (MSRD,Mrstn) Naval Reserve Group Com- THRESHOLD LOGIC techniques D. - alphanumeric, and other data in mand 4 -3(m) Camden, N.J.; 4/6/74 SOFTWARE DEVELOPMENT approach, Hempel (GCASD,Som) IEEE Intercon 1974, communications, computer, military, Real -time J.C. Kulp (MSRD,Mrstn) AFSC New York, Cont. Record, to ran- - Alternatives and other systems, CRT devices, Workshop on aeronautical systems software, dom logic Session 29/3; 3/26 -29/74 solid state displays, holographic dis- Dayton, Ohio; 4/3/74 plays, etc. 325 Checkout, Maintenance, and User Support NETWORK OPTIMIZATION program, 225 Antennas and Propagation Cascaded - V.K. Jha (RCA Ltd,Mont) Trans. MOVING MAP display, Holographic - B.R. automatic test equipment, (ATE), IEEE. Microwave Theory & Techniques, Vol. antenna design and performance; Clay (ATL,Cam) SPIE Conf. on Coherent maintenance and repair methods. 22, No. 3; 3/74

www.americanradiohistory.com Author Index Lockwood, H.F., 210 (ATL, Cam.) U.S. Pat. 3805170, April 16, 1974; Lurie, M., 345 Patents Assigned to U.S. Government Subject listed opposite each author's name Meier, E., 205 indicates where complete citation to his paper Mezrich, R.S., 360 Commercial Communications may be found In the subject index. An author Miller, A., 125 Granted Systems Division may have more than one paper for each Miller, R.E., 160 subject category. Mokross, B.J., 205 to RCA Engineers Automatic Squelch Tall Eliminator for Tone Napoli, L.S., 210 Coded Squelch Systems - L. F. Crowley, A. Astro- Electronics Division Narayan, S.Y., 215 M. Missenda, and D. R. Presky (CCSD, Nuese, C.J., 240 Meadowlands) U.S. Pat. 3810023, May 7, 1974 Maxwell, W., 225 Oka, T., 205 Okamoto, R., 205 Advanced Technology Phillips, W., 205 RCA Laboratories Laboratories Prager, H.J., 210 Priestly, E.B., 125 Consumer Electronics Low Beretringent Ortholerrites - R. B. Black O.D., 170 Raccah, P.N., 240 Clover, Jr. (Lab,. Pr.) U.S. Pat. 3804766, April Bocciarelli, B.B., 130 Reichert, W.F., 210 16, 1974 Brightness Control J. Avins (CE, Som.) Clay, B.R., 245 Robinson, P.H., 105 - U.S. Pat. 3804981, April 16, 1974 Corsover, S.L., 250 Rosen, J., 312 Semiconductor Memory Element - F. Dillon, L., 215 Ryan, R.J., 345 Sterzer (Labs, Pr.) U.S. Pat. 3805125, April 16, Information Playback System S. E. Hilliker Freiman, F.R., 175, 180 Sandercock, J.R., 205 - 1974 (CE, Indpls.) U.S. Pat. 3806668. April 23, 1974 Hansen, L., 240 Schade, H., 205 Lea, E., 240 Schiff, L., 215 Adaptive Surface Wave Devices A. Miller Static Convergence Device for Electron - O'Hara, L., 240 Shefer, J., 312 (Labs, Pr.) U.S. Pat. 3805195, April 16, 1974 Beams O. F. Thompson and J. L. Smith (CE, Richards, J., 360 Sigai, A.G., 240 - Indpls.) U.S. Pat. 3808570, April 30, 1974 Shelpuk, B., 280 Smith, R.T., 205 Least Recently Used Location Indicator - J. Sommers, H.S., 240 Shubert, W., 240 Television Receiver Using Synchronous A. Weisbecker (Labs. Pr.) U.S. Pat. 3806883, Staebler, D.L., 205 April 23, 1974 Video Detection J. Avins (CE, Som.) U.S. Staras, H., 312 - Commercial Communications Pat. 3812289, May 21, 1974 Systems Division Stewart, W.C., 210, 360 Method of Polishing Sapphire and Spinel - Swartz, G.A., 215 P. H. Robinson and R. O. Wance (Labs, Pr.) Independent Electron Gun Bias Control - J. Ben-Dov, 0., 225 Takahashi, T., 205 U.S. Pat. 3808065. April 30, 1974 C. Marsh, r. (CE. Indpls.) U.S. Pat. 3812397, Tietjen, J.J., 205 Bick, J.D., 250 May 21, 1974 Briel, L.A., 250 Turski, Z., 210, 215 Charge Transfer Fan -in Circuitry - P. K. Dickinson, RS., 250 Uphadhyayula, L.C., 215 Weimer (Labs, Pr.) U.S. Pat. 3811055, May 14, High Voltage Protection Circuit - M. N. Gluyas, T.M., 345 Wance, R.O. 105 1974 Norman (CE, Indpls.) U.S. Pat. 3813580, May J.L., Wargo, R., 210 Grever, 345 28, 1974 Sadashige, K., 250 Webb, G.W., 160 Aluminum Oxide Films for Electronic Devices Siukola, M., 225 Wehner, R.K., 240 Palm Beach Division - M. T. Duffy and J. E. Carnes (Labs, Pr.) U.S. Wen, C.P., 215 Pat. 3809574, May 7, 1974 Electronic Components Wettling, W., 205 Hinged Drum System - J. P. Watson (PBD, Wittke, J.P., 210 R. A. Palm Beach Gardens) U.S. Pat. 3813678, May Corona Discharge Device - Gange and E., 265 Williams, R., 125 Pr.) Pat. Baden, 28, 1974 C. C. Steinmetz (Labs, U.S. 3809974. Ehemann, G.M.. 210 Wolfe, M.E., 240 May 7, 1974 Hempel, D., 215 Woods, M.H., 125 Lindley, W.D., 210 Wronski, C.R., 210 Astro- Electronics Division Method of Treating a Glass Body to Provide Lyons, J.A., 180 Yocom, P.N., 205 an Ion -Depleted Region Therein - D. E. Dual Thrust Level Monopropellant Spacecraft Persyk, D.E., 210 Young, A., 215 Carlson, K. W. Hang, and G. F. Stockdale Propulsion System Y. C. Brilo (AED, Pr.) Rux, W.A., 365 Zaermowdki, T., 240 - (Labs. Pr.) U.S. Pat. 3811855. May 21, 1974 Zanzucchi, P.J., 205 U.S. Pat. 3807657. April 30. 1974 Government Communications & Method of Epitaxially Depositing Gallium Automated Systems Division RCA Limited Rotary Solenoid Shutter Drive Assembly and Nitride from the Liquid Phase- F. Z. Hawrylo Rotary Inertia Damper and Stop Plate and J. I. Pankove (Labs, Pr.) U.S. Pat. Farrell, A.T., 365 Conradi, J., 205, 210 Assembly - W. L. Cable and H. B Dougherty 3811983, May 21. 1974 Pr.) Pat. Fisher, C.D., 180 Feher, K., 340 (AED. U.S. 3804506. April 16, 1974; Fischer, H.L., 365 Jha, V.K., 370 Assianed to U.S. Government Two Color Medium for Full Color TV Film Kelly, G.E., 325 Shkarofsky, I.P., 125 System - R. E. Flory (Labs, Pr.) U.S. Pat. Spear, A.C., 175 Government Communication & 3812528, May 21, 1974 RCA Laboratories Automated Systems Division Missile & Surface Video Stripper - J. D. Cavett and R. S. Radar Division Abeles, B., 205, 210 Broad Band Single Frequency Generator - Hopkins, Jr. (Labs, Pr.) U.S. Pat. 3813488, Alig, R.C., 205 F. U. Everhard, Jr. (GCASD, Burl.) U.S. Pat. May 28. 1974 Adams, F.G., 315 Ametani, K., 210 3808548, April 30, 1974; Assigned to U.S. Buckley, M.W., 180 Appert, J.P. 205 Government Electronic Components Kulp, J.C., 365 Aries, Y., 210 Mercer, D.H., 170 Bartolini, R.A., 240 Processor Synchronization Scheme - R. M. Correcting Lens - A. M. Morrell and F. Van Schwartz, B.R., 170 Bolker, B.F.T., 205 Zieve, C. L. Maginniss and M. Kleidermacher Hekken (EC, Lanc.) U.S. Pat. 3811754, May Urkowitz, H., 215 Brunner, H., 205 (GCASD. Cam.) U.S. Pat. 3810119, May 7, 21, 1974 Hanak, J.J., 205, 210 Carnes, J.E., 210 1974; Assigned to U.S. Government Hannan, W.J., 345 Chiang, Y.S., 215 Method for Coating only the Conves Major Harada, S., 205 Clorfeine, A.S., 210 Automatic Positioning Device Exhibiting Surface of an Apertured Mask for a Cathode - Harbeke, G., 205 Cody, G.D., 205 High Accuracy and Repeatability - P. F. Ray Tube - B. K. Smith (EC, Lanc.) U.S. Pat. Harrington, W.C., 160 Cohen, R.W., 205 Minghella (GCASD, Burl.) U.S. Pat. 3811083, 3811926, May 21, 1974 Hicinbothem, W.A., 205 Coutts, M.D., 205 May 14, 1974; Assigned to U.S. Government Honig, R.E., 160 Crandall, R.S., 205 Impedance Control Using Transferred Elec- Huang, H.C., 210, 215 Curtis, B.J., 205 High -Speed Signal Following Circuit - N. R. tron Devices - R. E. Marx (EC, Pr.) U.S. Pat. Hughes, J.J., 210 Czaja, W., 205 Scheinberg (GCASD, Som.) U.S. Pat. 3812437, May 21, 1974 Hutchby, J.A., 205 Daniel, R.E., 210 3812383, May 21, 1974 Johnson, H.C., 312 deWolf, D.A., 130 Method for Making a Negative Effective Jolly, S., 210 Duncan, R.C., 205 Magnetic Reed Sensor Suitable for Use in Electron -Affinity Silicon Electron Emitter - Kaplan, G.S., 312 Enstrom, R.E., 205, 240 Ignition Timing Systems - L. R. Hulls A. H. Sommer (EC, Pr.) U.S. Pat. 3806372, Klensch, R.J., 312 Escher, J.S., 205 (GCASD, Burl.) U.S. Pat. 3813596, May 28, April 23, 1974 Knox, J.D., 210, 215 Ettenberg, M., 210, 240 1974 Kosonocky, W.F., 210 Feldstein, N., 170 Method of Installing a Mount Assembly in a Krausbauer, L., 205 Flory, R.E., 345 Government Engineering Multibeam Cathode -Ray Tube - J. F. Segro Kressel, H., 205, 210, 240 Gannon, J.J., 240 and G. L. Fassett (EC, Lanc.) U.S. Pat. Levine, P.A., 210 Halpern, T., 240 Transition Detector -- G. J. Dusheck, Jr. 3807006, April 30, 1974

www.americanradiohistory.com Method of Fabricating a Dark Heater - J. R. Method for Making a Radio Frequency Liquid Crystal Device Closure Method - A. Series Regulated Power Supply for Arc Dis- Hale and G. I. Merritt (EC, Lanc.) U.S. Pat. Transistor Structure - D. S. Jacobson and R. N. Gardiner and H. A. Stern (SSD, Som.) U.S. charge Lamps Utilizing Incandescent Lamps A. Duclos 3808043, April 30. 1974 (SSD. Som.) U.S. Pat. 3807039, Pat, 3808769, May 7. 1974 - F. Caprari (SSD, Som.) U.S. Pat. 33813576, April 30, 1974 May 28, 1974 Solid State Division Liquid Crystal Display - R. C. Heuner and S. Method of Closing a Liquid Crystal Device - J. Niemiec (SSD, Som.) U.S. Pat. 3809458, Current Source- A. C. N. Sheng (SSD, Som.) Hybrid Electron Device Containing Semicon- H. A. Stern (SSD, Som.) U.S. Pat. 3807127, May 7, 1974 U.S. Pat. 3813595, May 28, 1974 ductor Chips - E. T. Hausman (SSD, Som.) April 30. 1974 U.S. Pat. 3805117, April 16, 1974 Window Detector Circuit - A. W. Young Record Division Capacitive- Discharge Timing Circuit Using (SSD, Som.) U.S. Pat. 3809926, May 7, 1974 Gated Astable Multivibrator - A. J. Visioli. Jr. Comparator Transistor Base Current to Tape Cartridge Player Cartridge Magazine - and H. A. Wittlinger (SSD. Som.) U.S. Pat. Determine Discharge Rate - L. R. Campbell Set -Reset Flip -Flop - G. E. Skorup (SSD, B. A. Grae and D. R. Andrews (Record Div., 3805184. April 16, 1974 (SSD, Som ) U.S. Pat 3808466, April 30, 1974 Cam.) U.S. Pat. 3812384, May 21, 1974 Indpls.) U.S. Pat. 3812537, May 21, 1974

Engineering News and Highlights

the PhD from Columbia University in 1911. America, the Optical Society of America, He was given a lifetime appointment at and the International College of Surgeons. CCNY as Associate Professor of Electrical Engineering. He was also granted an Among his many honors, Dr. Goldsmith Honorary Doctor of Science degree by was the recipient of the IRE Medal of Lawrence College in 1935. Honor in 1941, and the IRE Founders Award in 1954. In March 1962, he was Dr. Goldsmith was consulting engineer for honored with a scroll of tribute by the IRE the General Electric Company from 1915 at its Golden Anniversary Celebration. Dr. to 1917, and Director of Research for Goldsmith also received the Progress Marconi Wireless Telegraph Company of Medal Award of the Society of Motion America from 1917 to 1919. Dr. Goldsmith Picture and Television Engineers; Modern joined RCA in 1919 when it acquired the Pioneer Award (1940); Townsend Harris Marconi Company. At the start, Dr. Medal (1942). Goldsmith and his staff conducted their research and engineering efforts at the Dr. Goldsmith was elected President of the Electrical Engineering Laboratory of Society of Motion Picture and Television CCNY. As the work load increased RCA Engineers in 1932, and had been chairman established a Technical and Test Depart- and member of various sectional comment at Van Cortlandt Park, in New York mittees of the American Standards City, which became the Company's first Association. In 1966, he became Vice research laboratory. President (Electronics) of the Pan - Dr. Alfred N. Goldsmith American Medical Association. He was an (September 15, 1888 -July 1, 1974) In 1920, Dr. Goldsmith's work made possi- Honorary member of the Radio Society of ble the first commercial radio with only Great Britain, the Royal Society of Arts Dr. Alfred N. Goldsmith, a world - two control knobs and a built -in speaker; (Great Britain), the International Com- renowned scientist, engineer and inven- and also the first commercial radio - mittee of Radio Telegraphy (France), the tor, died July 1, in St. Petersburg, Fla., phonograph. He also made vital con- Academy of Motion Picture Arts and where he had resided for the past 16 tributions to the development of the first Sciences, the Society of Motion Picture months. He was 86. color television tube to find commercial and Television Engineers, and of the In- and worldwide use. He proposed a color stitution of Radio Engineers (Australia). Dr. Goldsmith was an Honorary Vice television picture tube employing a screen He was also an eminent member of Eta President of RCA and its Senior Technical of color phosphor dots and a perforated Kappa Nu Association, and a member of Advisor at the time of his death. A prolific plate. In its simplest terms this was the the New York Medico -Surgical Society. inventor, he was a pioneer in the develop- basic idea of the shadow -mask color pic- ment of radio and television, sound motion ture tube now in widespread commercial pictures, medical electronics and other use throughout the world. areas of electronics. Jellinek receives Thurlow Navigation When RCA was formed in 1919, the late Dr. Goldsmith was a Co- Founder, Fellow, Award General David Sarnoff named Dr. Director, and Life Member of the Institute Goldsmith to head the new company's of Radio Engineers, was its President in Ernest Jellinek was honored by the In- research and development activities. Dr. 1928: and had served as Editor, Editor stitute of Navigation for making the most Goldsmith served as Director of Research Emeritus, and Anniversary Editor since its outstanding contribution to the science of and later as Vice President and General founding in 1912. He was Secretary of the navigation during 1973. Engineer of RCA from 1919 to 1931, when IRE from 1918 to 1927. He was also a he resigned to become an independent Fellow of the American Institute of Elec- Mr. Jellinek receieved the Thomas L. consultant. He remained as Senior trical Engineers. In 1963, the IRE and AIEE Thurlow Navigation Award at a banquet Technical Consultant to RCA. merged to form the IEEE, and Dr. held during the Institute's annual meeting Goldsmith became a Fellow, and Director in San Diego, Cal. Mr. Jellinek is Staff Born in New York City, September 15, and Editor Emeritus of the IEEE. Systems Manager, Navigation and Traffic 1888, Dr. Goldsmith began his career as an Control, Government Plans and Systems instructor at the College of the City of New In addition, Dr. Goldsmith was a Fellow of Development, for RCA Government and York in 1907, when he also received the BS the American Physical Society, the Commercial Systems, Camden, N.J. from that college. He remained with CCNY American Association for the Advance- until 1919 and, in the meantime, received ment of Science, the Acoustical Society of Mr. Jellinek was cited for his exceptional

www.americanradiohistory.com ingenuity and engineering skill in conceiv- Program for all of the Government and National Association for Remotely Piloted ing and designing RCA's Harbor Traffic Commercial Systems Divisions involved in Vehicles (RPVs). Ranging, Identification and Communica- government contracts - administering tion (HATRIC) System. (This system was the fields of investigation and the The Association, with headquarters in described in the April -May 1973 issue of allocations of funds for the program. Dayton, Oh o, was organized in 1972 for the RCA Engineer.) the "advancement of remotely employed During Mr. Cottler's tenure as Chief air power to complement manned aircraft Mr. Jellinek joined RCA in 1958 as a Engineer of the Missile and Surface Radar forces." Project Engineer with the company's Division (1967- 1974), the Division made Communications Systems Division. He important advances in radar and systems Mr. Shore earned the BS in Aeronautical was a leader in programs to develop technology, ranging from such major Engineering from the University of Jellinek integrated navigation and com- systems developments as AEGIS and Michigan in 1941 and the MS in Physics munications systems for advanced aircraft AN /FPS -95 to such critical concept from Ohio State University in 1950. such as the B -1 and AWACS. developments as ultra- high -speed digital logic design for advanced data /signal Mr. Shore has been with the company In 1968, he joined the Systems Develop- processing systems. since 1954. Presently, he directs the ment Group where he is technical director planning activities for G &CS and the for advanced systems development in the Mr. Cottler received the BChE from the development of major new weapons area of navigation and traffic control for College of the City of New York in 1942 systems concepts. Mr. Shore has led aircraft and ships. Prior to joining RCA, and later took graduate courses in elec- G &CS' multi -divisional efforts in several Mr. Jellinek worked for General Electric trical engineering at Polytechnic Institute RCA and Air Force programs involving and the Electronics Corporation of of Brooklyn as well as management RPVs, including the Multi- Mission RPV America. courses while with the Government and and Drone Control and Data Relay Volpe RCA. System. Mr. Jellinek earned the BEE from the Polytechnic Institute of New York in 1941. Mr. Cottler was Project Manager for the Presently, Mr. Shore is Chairman of the He is a senior member of IEEE and holds BMEWS Tracking Radar (AN /FPS -49) NSIA Command, Control and Corn - memberships in the Institute of Navigation during its design, development, produc- munications Advisory Committee and was and the American Institute of Aeronautics tion, and test and later became the Chairman of the IEEE International Con- and Astronautics. He is a Licensed BMEWS engineering manager during im- ference on Communications in 1967. He is Professional Engineer in the state of New plementation of the United Kingdom site. a member of several professional York. He joined RCA in 1956 where his initial associations and holds a Professional responsibility was Project Leader for Engineering License for the state of New tracking-radar development and field Jersey. evaluation in the Talos Land -Based Mis- Cottler Volpe named Chief Engineer at MSRD sile Program. Previous to this he spent eight years at White Sands Proving Max Lehrer, Division Vice President and Ground beginning in 1948. He was also a General Manager, Missile and Surface charter member of the Electronic Trajec- Amantea and Held named David Sarnoff Radar Division, Moorestown, N.J., recent- tory Measurements Working Fellows ly appointed Joseph Group of the C. Volpe as Chief Inter -Range Engineer. Instrumentation Group (IRIG). Dr. William M. Webster, Vice President, RCA Laboratories, Princeton, N.J., recent- In his new post, Mr. Volpe will act as the He is a member of the AIAA, American ly announced that two Solid State principal engineering executive of the Ordnance Association and Air Force Technology Center. scientists have been Missile and Surface Radar Division with Association. awarded David Sarnoff Fellowships for responsibility for all engineering activities. graduate study in the 1974 -75 academic Shore He also will be responsible for develop- year. ment of advanced techniques needed for Rabinowitz appointed Principal Scientist future systems. One of the Fellows, Robert Amantea, plans to attend the Massachusetts Insitute of Max Lehrer, Division Vice President and Mr. Volpe received the BS in physics from Technology and work toward the PhD in General Manager, Missile and Surface St. Joseph's College and completed Electrical Engineering. Gerald D. Held, Radar Division, Moorestown, N.J., ap- graduate courses at Temple University who is working toward the PhD in Com- pointed Dr. Samuel J. Rabinowitz as and the University of Pennsylvania in puter Science, will start his Principal Scientist. second year as physics and servomechanisms. He is a a Sarnoff Fellow at the University of member of Sigma, the honorary physics California, Berkeley. fraternity. In the newly- created post, Dr. Rabinowitz will act as a technical adviser to Mr. Lehrer The David Sarnoff Fellowships, es- Amantea on the various radar, data processing and Prior to his promotion, Mr. Volpe was tablished in 1956 to commemorate the late related systems under development at the Project Manager for the AEGIS Weapon General Sarnoff's fifty years in Division. He also will radio, System being developed by RCA for the perform special assignments. television and electronics, are awarded Naval Ordance Systems Command. annually to outstanding employees of RCA. The Fellowships are the top Prior to his promotion, Dr. Rabinowitz, as Since joining RCA in 1958, Mr. Volpe has educational awards available to RCA Manager of Systems Engineering, was held increasingly responsible managerial employees. posts in various radar development responsible for directing the Division's efforts to develop new systems to meet programs at the Moorestown facility. Mr. customer needs. Amantea has worked for RCA since receiving the BEE in 1965 from the City University of New York. He was awarded Cottler on G &CS Staff the MS in Electrophysics in 1969 from the Held Dave Shore elected President, National Polytechnic Institute of New York. Dr. Harry J. Woll, Vice President, Govern- Association for Remotely Piloted Vehicles ment Engineering announced the ap- Mr. Held, an RCA employee since 1970, pointment of Dudley M. Cottler as Staff David Shore, Division Vice Presicent, received the BS in Electrical Engineering Technical Advisor. Mr. Cottler's primary Government Plans and Systems Develop- from Purdue University in 1970, and the responsibility in his new position is the ment, Government and Commercial MSE in 1972 from the University of Penn- Independent Research and Development Systems, has been elected President of the sylvania.

www.americanradiohistory.com New appointments at Consumer Elec- of Ships. Mr. Wright was transferred from He received an RCA Laboratories Out- tronics the Radio Design Group to Resident standing Achievement Award in 1967. Two Engineering at the Bloomington Plant and years later, he became the first person in Loren R. Kirkwood, Division Vice Presi- became Resident Engineer in 1951. He the history of RCA Laboratories to receive dent, Television Engineering and was responsible for the coordination of two Outstanding Achievement Awards in Strategic Planning, announced the Engineering activities with the Manufac- one year for work on separate research organization of Television Engineering turing Plant in the areas for radio, black tems. and Strategic Planning as follows: and white tv, color tv and Selectavision. Jobe Thornley C. Jobe, Chief Product Develop- In 1973, he was given RCA's highest ment Engineer - Color Television; Robert technical honor, the David Sarnoff Award, J. Lewis, Chief Engineer, Black and White Williams invited to serve on for outstanding team research leading to Television Product Design and Safety Scientific Advisory Committee a new class of integrated semiconductor Engineering; James A. McDonald, arrays." Manager, Technology and Services; and Dr. John C. Williams Staff Engineer, John M. Wright, Chief Product Design Government Plans and Systems Develop- He is a member of Sigma Xi and the Engineer - Color Television. ment, Camden, N.J., has been invited to Electrochemical Society as well as being a serve as an Associate Member of the senior member of the IEEE. Mr. Scott has Thornley Jobe received the BSEE from Defense Intelligence Agency's Scientific been issued 14 U.S. Patents and has Georgia Institute of Technology. He Advisory Committee. written about 25 technical papers. ItLewis joined RCA in 1953 as Liaison Engineer at The Scientific Advisory Committee the Home Instruments Division (now Con- Schindler to head IEEE Section sumer Electronics) television receiver provides a valuable link between the plant Bloomington, In- Defense Intelligence Agency and the manufacturing Dr. Max J. Schindler of the Microwave 1961, he became Leader in the scientific and industrial communities of diana. In Device Operations Department, Elec- Television Design section at In- the nation. Dr. Williams was nominated by Color tronic Components, Harrison, N.J., has dianapolis, Indiana. In 1966, he became Vice Admiral V.P. dePoix, USN, Director, been elected Chairman of the North Manager of Resident Engineering at the Defense Intelligence Agency, and subse- Jersey Section of the Institute of Electrical Memphis, Tennessee, television receiver quently approved by the Office of the and Electronics Engineers. The IEEE is the plant. In 1971, he became Manager of Secretary of Defense for appointment as world's largest Engineering Society, with a Product Analysis and Control at In- an Associate Member for a one -year term world -wide membership of over 170,000, dianapolis. He assumed his present posi- beginning July 1, 1974, subject to possible of which 4,500 reside in the North Jersey tion in 1973. He is a member of Eta Kappa extension. McDonald area. Nu and Tau Beta Pi. He has been awarded Dr Williams received the BS in Electrical five U.S. patents. the MSEE and the Engineering from Pennsylvania State Dr. Schindler received of Technical Sciences from the College in 1941; the MS in Engineering Doctor in Vienna, Austria. Science and Applied Physics from Har- Technical University Robert Lewis received the BSEE from in 1958 and has worked on vard in 1948; and the PhD in Applied He joined RCA Drexel Institute of Technology. He started technical problems Physics from Harvard in 1959. He joined a number of basic with RCA at Camden, N.J. in 1941 as an traveling -wave Government Plans and Systems Develop- related to the design of electrical engineer. When the RCA Home -field devices, ment in 1968 where he has provided tubes, magnetrons, crossed Instruments Division moved to In- has technical direction and management fora and solid state devices. Dr. Schindler dianapolis, Ind. from New Jersey, to numerdus number of electromagnetic systems published and presented become RCA Consumer Electronics he programs. papers and holds three patents. He is a Wright was responsible for organizing and train- senior member of IEEE. In accepting his ing an engineering group at the new home Dr. Williams is a member of the American new post, Dr. Schindler commented, "Our office of the division. He was named Physical Society, American Geophysical Section will endeavor to establish closer Manager, Black- and -White TV Engineer- Union, New York Academy of Sciences, ties to the local communities. Engineers ing in 1969 and in 1973 became Chief and the American Association for Ad- have developed the technology which nas with engineering design Engineer vancement of Science. made this country the richest in the world, responsibility for all RCA black- and -white and they are able and anxious to solve the tv receivers made in domestic and foreign environmental and socio- economic side plants. Scott appointed Director of IC effects of this high technology." Technology Research BSC James McDonald received the granted (honors) from Glasgow University in 1955 Gerald B. Herzog, Staff Vice President, Degress Williams and the MS from Purdue University in Technology Centers, at RCA Laboratories M. Vaughan of Electronic 1968. He worked for Associated Electrical in Princeton, N.J., recently appointed Steven Recording Equipment Engineering, Com- Industries in England for seven years, Joseph H. Scott, Jr., as Director of In- Systems Divi- mainly on signal processing for radar. In tegrated Circuit Technology. mercial Communications 1962, Mr. McDonald joined RCA Con- sion, Camden, N.J., received the Master of sumer Electronics and designed deflec- Mr. Scott received the AB in Chemistry Science in Systems Engineering from the tion circuits for Black and White Televi- from Lincoln University in 1957. The University of Pennsylvania. sion for which he received twelve patents. following year he attended the Graduate In 1968 he was promoted to Leader in School of Chemistry at Howard Universi- Edward Goldman, Manager of Microwave Advanced Development, and his ty. He also did graduate work in Electrical Purchasing, Electronic Components, assignments included a video tape player, Engineering at Howard after joining RCA. Harrison, N.J. received the Master of a holographic player, and a single -tube Science in Business Administration from to Scott color camera. In 1973 he was made Mr. Scott has been with RCA since 1959. Rutgers University. He also was elected manager of Video Disc Engineering. He started with Electronic Components the the Beta Gamma Sigma Honorary and Devices in Somerville, N.J., doing Business Fraternity. John Wright received the BA in Math and research and development work on Physics from Wabash College in 1938 and semiconductor devices. He transferred to Thomas L. Nevue of the Solid State later attended both George Washington RCA Laboratories in 1967. In 1970, he was Production Engineering Group, University and Indiana University. Prior to named Head of the Integrated Circuit Microwave Devices Operating Depart- joining RCA in Camden in 1946, Mr. Technology and Applications Research ment, Electronic Components, Harrison, Wright was engaged in radio receiver Group, the position he held until his N.J., received the Bachelor of Science in design with the Navy Department, Bureau promotion. Electrical Engineering from Newark

Schindler

www.americanradiohistory.com College of Engineering. Professional activities Killion delivered papers at the Fifth Annual Reliability Technology Seminar. Menzo A. Frank J. Strobl, Administrator, Technical RCA Laboratories Christman, Manager, Assembly Quality Communications, Engineering Profes- Control, is Editor of the monthly ASQC Programs, sional received the Bachelor of Dr. Jan A. Rajchman, Staff Vice President, Newsletter. Science in Management from Rutgers Information Sciences was elected a Fellow University. of the American Physical Society.

William Shibler of Aural Broadcast Tom Cook, DSRC photographer, was Engineering, Communications Systems cited by Industrial Photography magazine Government Communications and Division, Meadow Lands, Pa., received the for his photo of a holographic image, Automated Systems Division Master of Science in Electrical Engineer- "considered to demonstrate the most out- Burlington Operations ing from the University of Pittsburgh. standing level of technical achievement of all photographs submitted" in the Veronica Hsu was elected Secretary of the Andrew C. Billie, Jr., of Aural and TV Industrial Photography 1974 Annual. Aerospace Division of the Special Broadcast Publications, Communications Libraries Association. Also, she served as Systems Division, Meadow Lands, Pa., Chairman of the Sci Tech Committee, received the Bachelor of Arts in Boston Chapter, for two years. Mathematics from Washington and Jeffer- Missile and son College. Surface Radar Division Bill Curtis was elected by the Board of Directors of the Electrical Engineering Paul Rappaport, Director of the Process Howard Wintling, Manager, Material Alumni Association of the University of and Applied Materials Research Quality Control, was elected Second Vice Illinois, Urbana, Ill., as one of three Laboratory at RCA Laboratories was Chairman of the Philadelphia Section recipients of the Distinguished Alumnus awarded an Honorary Doctor of Science American Society for Quality Control (AS- Award in recognition of his outstanding Degree from Arizona State University in QC). George J. Brannin, Manager, contributions to the Electrical Engineer- recognition of his pioneering research on Product Assurance, served as moderator ing professlion and for his long- standing solar energy and development of the solar of the ASQC May dinner meeting - over loyal support of Electrical Engineering at cell. 100 attended. Fulvio Olivetto and Robert Illinois.

Promotions sion Operations (J. Christopher, New R. Shaver from Mgr., Overseas Telex Pro- York) jects to Mgr., Iran ICI Project (A. A. Avanessians, New York) Global Communications, Inc. Jack L. Ray from Mgr., Data Networks to Mgr., Applications Engineering (A. W. Electronic Components A. Annibell from Mgr., Computer Systems Brook, New York) Sales to Mgr., Videovoice (P. Schneider, C. E. Dover from Sr. Eng. Prod. Dev. to New York) A. A. Avanessians from Dir., Computer Engineering Projects to Dir., Advanced Eng. Ldr. Prod. Dev. (T. E. Yingst, Lan- caster) J. Christopher from Mgr., Spacecraft Switching Projects (P. Schneider, New Dir., York) Engineering to Satcom Project (P. K. Wright from Engr., Mfg. to Mgr., Schneider, New York) L. Correard from Mgr., Computer Production Engrg. (N. Meena, Marion) J. J. Dietz from Engrg. Ldr. to Mgr., Switching Engineering to Mgr., Advanced Telex Engineering (A. A. Avanessians, D. N. Herd from Engr., Mfg. to Mgr., Digital /Voice Design Engineering (A. A. Production Engrg. (N. Meena, Marion) Avanessians, New York) New York) J. Cuddihy from Mgr., Satellite Engrg. to E. L. Batz from Ldr., Liaison Engr., Resi- B. Maier from Engr. to Group Ldr., Mgr., Earth Station Engrg. (J. JM. Walsh, dent Engrg. to Mgr., Resident Engineering Overseas Telex Projects (R. Shaver, New (J. M. Wright, Bloomington) York) New York) Solid State Division M. Rosenthal from Mgr., Engineering J. Gleitman from Mgr., Computer Systems Engrg. to Mgr., Telex Switching Marketing Services, to Mgr., Administration and R. Rhodes from Member, Tech. Staff to Planning (P. New (A. A. Avanessians, New York) Schneider, York) Ldr., Tech. Staff (E. P. Zlock, Som.) D. L. Lundgren from Mgr., Project S. Schadoff from Mgr., Commercial Administration to Mgr., Program Control A. Dingwall from Sr. Member, Tech. Staff Leased Channels to Dir., Commercial (A. W. Brook, New York) to Ldr., Tech. Staff (I. H. Kalish, Som.) Leased Channels Engineering (V. Ar- bogast, New York) L. Ottenberg from Mgr., Telecom- munications Systems to Mgr., Com- RCA Alaska Communications, Inc. L. J. Wilson from Mgr, Computer Systems munications Systems Engrg. (A. W. Brook, to Dir., Computer Systems (P. Schneider, New York) E. R. Leavens from Real Estate Specialist New York) to Mgr., Real Estate (K. J. Rourke, D. Mandato from Mgr., Special Equipment Anchorage) M. ChaFong from Mgr., Message Engineering to Mgr., Special Equipment Switching Engineering to Mgr., Computer Design Engineering (A. A. Avanessians, G. V. Bartley from Mgr., Logistics Engineering (L. J. Wilson, New York) New York) Coordination to Mgr., Engineering Administration (G. P. Roberts, Dr. Armand DilPare from Mgr., Launch J. McDonald from Mgr., Telex Engineering Anchorage) Vehicle and Mission Reports to Mgr., to Mgr., Services Engrg. (R. J. Angliss, Launch Vehicle (J. Christopher, New New York) Astro- Electronics Division York) W. W. Schaefer from Mgr., Telex Quality J. J. Hawley from Sr. Engr. to Mgr. Corn - C. Lundstedt from Mgr., Earth Station and Control & Planning to Mgr., Traffic munications Systems (A. Aukstikalnis, Data Subsystems to Mgr., TTC and Mis- Engineering (R. J. Angliss, New York) AED, Pr.)

www.americanradiohistory.com Beckmann appointed to Corporate Service Company Engineering Staff Julius Koppelman, President, RCA Service Company has announced the ap- Howard Rosenthal, Staff Vice President, pointment of George G. Ray as Managing Engineering, at the RCA David Sarnoff Director, Service Division, RCA Limited Research Center, Princeton, N.J., ap- (U.K.) pointed Gerald K. Beckmann to the RCA Corporate Engineering Staff. Parts and Accessories Mr. Beckmann was graduated from the William B. Buchanan, Manager, Product Virginia Polytechnic Institute with the BS Barton Birmingham Operations has announced the appoint- in Electrical Engineering in 1964. He has ment of Robert A. Famiglietti as Manager, done graduate work in Engineering Product Handling. at Northeastern University. Management Barton is new TPA for Mr. worked for the Honeywell Beckmann Electronic Components Mobile Communications and the IBM Corporations prior to joining In 1973 Mr. Beckmann received Plant Manager, Lan- RCA. Richard H. Hynicka, Frederick A. Barton has been appointed Honorable Mention in the Eta Kappa Nu caster Color, Picture Tube Plant has an- Technical Publications Administrator for Outstanding Young Engineer competi- nounced the appointment of William J. Mobile Communications Systems, Com- tion. He was cited for "his notable ac- Harrington as Manager, Manufacturing. mercial Communications Systems Divi- in computer software and complishments sion, Meadowlands, Pa. In this capacity, his involvement in his community." D. Manager, Electro- Eugene Savoye, Mr. Barton is responsible for the review Optics Product Development Engineer- and approval of technical papers; for ing, Industrial Tube Division, has an- coordinating the technical reporting nounced the Electro- Optics Product program; and for promoting the prepara- Staff announcements Development organization as follows: tion of papers for the RCA Engineer and Thomas W. Edwards, Manager, Material & other journals, both internal and external. Process Development; Fred A. Helvy, President and Chief Operating Officer Manager, Product Development Mr. Barton received an HNCEE degree Engineering; Frederick R. Hughes, Anthony L. Conrad, President and Chief from the Borough Polytechnic, London, Manager, Solid State Emitters Develop- Operating Officer has announced the England in 1944. He served as Radar ment; and Donal C. Reed, Manager, election of Joseph W. Curran as Vice Officer with the Royal Indian Navy from Custom Product Engineering. President of the Corporation. 1945 to 1947. From 1948 until 1955 he was a development engineer in the Com- Robert C. Demmy, Manager, Product New Business Programs munications Department of Redifon Ltd., Engineering, Scranton Plant, has an- Development London. After two years with Central Corporate the as follows: nounced organization Rediffusion Services as a CATV Systems George T. Aschenbrenner, Engineering Richard W. Sonnenfeldt, Acting General Engineer in the field, he rejoined Redifon Leader, Applications Lab.; Louis J. Manager, Electronic Industrial Engineer- Ltd. as Section Leader in charge of HF DiMattio, Engineering Leader, Design ing Division has announced the organiza- Transmitter Design. In 1960, he joined the Lab.; and Robert J. Murray; Engineering tion as follows: Henry Duszak, Operations Mobile Communications Engineering Leader, Chemical & Physical Lab. Manager; Henry Duszak, Acting Department of RCA and has had design Marketing; John L. Ovnick, Chief and management responsibility for a Solid State Division Engineer; Robert L. Schoenbeck, Staff variety of mobile products. A Senior Engineer in the Advanced Development Technical Advisor; Peter A. Weston, Bernard V. Vonderschmitt, Vice President Department, Mr. Barton became leader of Director, Finance; and James F. Wiard, and General Manager has announced the this group in April, 1974. Plant Manager. organization of the Solid State Division:

Robert M. Cohen, Director , Quality and Astro- Electronics Division Reliability Assurance; Walter B. Dennen, Birmingham is new TPA for GPSD Manager, News and Intormation; D. Mark Sasso, Manager, Plant Operations Joseph Donahue, Division Vice President, James T. Birmingham has been appointed has announced the apointment of Leo Solid State International Operations; Technical Publications Administrator for Weinreb as Manager, Manufacturing. Edward K. Garrett, Division Vice Presi- Government Plans and Systems Develop- dent, Finance; Ben A. Jacoby, Division ment, Camden, N.J. In this capacity, Mr. Missile and Surface Radar Division Vice President, Solid State Marketing; Birmingham is responsible for the review Donald W. Ponturo, Division Vice Presi- and approval of technical papers; for Max Lehrer, Division Vice President and dent, Industrial Relations; Richard A. coordinating the technical reporting General Manager has announced the ap- Santilli, Division Vice President, Solid program; and for promoting the prepara- pointment of Lewis Nelson as Manager, State Bipolar Integrated Circuits and tion of papers for the RCA Engineer and Strategic Systems Department. Special Products; Edward M. Troy, other journals, both internal and external. Division Vice President, Solid State Services and Offshore Manufacturing; Mr. Birmingham received the BS in In- Government Communications and Carl R. Turner, Division Vice President, dustrial Management from LaSalle Automated Systems Division Solid State Power Devices; and Harry College in 1960 and has completed Weisberg, Division Vice President, State specialized courses in Government Con- James M. Osborne, Division Vice Presi- MOS Integrated Circuits. tract Administration at Harbridge House dent and General Manager has an- Inc. and William and Mary College. In nounced the appointment of Charles G. D. Joseph Donahue, Division Vice Presi- 1962, he joined the Missile and Surface Arnold as Manager, Commercial dent, Solid State -International Radar Division, as an administrator on the Telecommunications Systems and an- Operations has announced the Inter- BMEWS Program and in 1964 transferred nounces the organization as follows: national Operations organization as to the Government Plans and Systems Philip A. Gehman, Manager, Telex Equip- follows: Joseph W. Karoly, Division Vice Development activity where he is present- ment and Systems; Raymond J. Kowalski, President, Solid State - Europe; Robert L. ly Administrator, Engineering Administra- Manager, PABX Equipment and Systems; Klem, Manager, International Sales; and tion. Mr. Birmingham, in his current and William G. Herold, Administrator, Pro- Robert E. O'Brien, Administrator, Euro- assignment, is responsible for all phases ject Control. pean Operations Support. of program administration.

www.americanradiohistory.com Editorial Representatives The ï`.ii'= .nal t e presf r:tative :n ynur rtrr ,E, a 'nr ,nr' Y u s ,r,n(. iuhng te ,nn.c.a' [írß, ar)C . "n0 Government and Commercial Systems Astro- Electronics Division t. M. SEIDEMAN' Engineering, Princeton, N.J. y. WEISBERN'' Advanced Development and Research, Princeton, N.J. Commercial Communications Systems Division Broadcast Systems N N HURST Broadcast Systems Advanced Development, Camden, N.J. R. E. WINN Broadcast Systems Antenna Equip. Eng., Gibbsboro, N J. A. C. BILLIE Broadcast Engineering, Meadowlands, Pa. Mobile Communications Systems F. A. BARTON Advanced Development, Meadow Lands, Pa.

Electromagnetic and C. S. METCHETTE' Engineering, Van Nuys, Calif. Aviation Systems Division J. McDONOUGH Aviation Equipment Engineering, Van Nuys, Calif.

Government Communications and A. LIGUORI' Engineering, Camden, N.J. Automated Systems Division K. E. PALM' Engineering, Burlington, Mass.

Government Engineering M. G. PIETZ' Advanced Technology Laboratories, Camden, N.J. J. E. FRIEDMAN Advanced Technology Laboratories, Camden, N.J. J. L. KRAGF t Central Engineering, Camden, N.J.

Government Plans and E. J. PODEL=. Eng. Information and Communications, Camden, N.J. Systems Development

Missile and Surface Radar Division D. R. HIGGS _ Engineering, Moorestown, N.J.

Palm Beach Division B. B. BALLARD' Palm Beach Gardens, Fla. Research and Engineerini, Laboratories C. W. SALL' Research, Princeton, N.J. I. H. KALISH Solid State Technology Center, Somerville, N.J. M. R. SHERMAN Solid State Technology Center, Somerville, N.J.

Electronic Components C.A. MEYER' Chairman, Editorial Board, Harrison, N.J. Entertainment Tube Division J. KOFF Receiving Tube Operations, Woodbridge, N.J. J. H. LIPSCOMBE Television Picture Tube Operations, Marion, Ind. E. K. MADENFORC' Engineering, Lancaster, Pa.

Industrial Tube Division J. M. FORMAN Industrial Tube Operations, Lancaster, Pa. H. J. WOLKSTEIN. Solid State Product Development Engineering, Harrison, N.J Consumer and Solid State Electronics Solid State Division E. M. McELWEE Chairman, Editorial Board, Somerville, N.J. J. E. SCHOEN Solid State Division, Somerville, N.J. J. DiMAURO Solid State Division, Mountaintop, Pa. S. SILVERST, Power Transistors, Somerville, N.J. H. A. UHL Integrated Circuits, Somerville, N.J. J. D. YOUNC Solid State Division, Findlay, Ohio

Consumer Electronics C. W. HOYT Chairman, Editorial Board, Indianapolis, Ind. R. BUTH Engineering, Indianapolis, Ind. R. C. GRAHA'. Audio Products Engineering. Indianapolis, Ind. F. HOLT Advanced Development, Indianapolis, Ind. E. E. JANSC. Black and White TV Engineering, Indianapolis, Ind. J. STARK Color TV Engineering, Indianapolis, Ind. P. HUANG Engineering, RCA Taiwan Ltd., Taipei, Taiwan Services RCA Service Company M. G. GANDER` Consumer Services Administration, Cherry Hill, N.J. W. W. COOK Consumer Service Field Operations, Cherry Hill, N.J.

R. M. DOMBROS', - Technical Support, Cherry Hill, N.J. R. I. COGHILL Missile Test Project, Cape Kennedy, Fla.

Parts and Accessories C. C. REARICK' Product Development Engineering, Deptford, N.J.

RCA Global Communications, Inc. W. s. LEIS* RCA Global Communications, Inc.,New York, N.Y. P. WEST* RCA Alaska Communications, Inc., Anchorage, Alaska

NBC, Inc. W. A. HOWARD' Staff Eng., Technical Development, New York, N.Y.

RCA Records M. L. WHITEHURS1 Record Eng., Indianapolis, Ind.

Corporate Development C. A. PASSAVANT' International Planning, New York, N.Y.

RCA Ltd W. A. CHISHOLM* Research & Eng. Montreal, Canada

Patent Operations M. S. WINTERS Patent Plans and Services, Princeton, N.J.

*Technical Publications Administrators (asterisked ' above) are responsible for review and approval of papers and presentations

www.americanradiohistory.com Engineer

A TECHNICAL JOURNAL PUBLISHED BY RESEARCH AND ENGINEERING "BY AND FOR THE RCA ENGINEER"

Printed in U.S.A. Form No. RE -20 -1