Edr‘r. Note: In response to requests from many film workers and others not yet inducted into
`the intricacies of digital television. the following introduction has been prepared by David A.
`Howell of the Society‘s Editorial Staff, with substantial parts and suggestions contributed by
`many members ofthe Board of Editors. This primer provides background information. the
`philosophy and fundamental science of digital television. but little of the engineering. The
`next paper, by B. S. Busby. is also a digital television background paper. but here the empha-
`sis is on showing the principles from an engineering point of view. i.e. it is shown how televi-
`sion can be digitized by using analog-to-digital and digital-to-analog converters and other
`equipment.
`It is recommended that the reader examine both papers and select the approach most suit-
`able for him. Digital television is topical. challenging, abstruse and highly technical. We hope
`that. having the advantage of two background papers, many more Journal readers will benefit
`from the digital television papers that are sure to come in the future.
`
`ic signal is used as an "analogy" to rep-
`
`Digital Television
`
`A Primer on Digital Television
`
`By DAVID A. HOWELL
`
`ment of change in the variable output of
`a device. The resolution of any analog
`device depends on the exactness of the
`analogy that is used. on a factor which
`relates the magnitude of a number to the
`precision of its representation by the de-
`vice (the scale famor) and to some extent
`on the estimating skill of the operator.
`The resolution of the readout of a digital
`device depends exclusively on the num-
`ber of significant figures or "places" that
`one is willing to pay for. (Four and six-
`digit calculators are quite cheap. while
`twelve-digit ones are more expensive —-
`and more accurate than most people
`need.) The accuracy of the readout ob-
`tained with a digital device in no way de-
`pends on the operator's estimate.
`We shall see later in this paper that
`when it comes to digital television. extra
`digits are very important but also very
`expensive in terms of money. complexity
`and something called bandwidth.
`
`Conventional Television in Brief
`
`Conventional television is analog tele-
`vision. When a scene is scanned. the cur-
`rent
`from the cathode of a
`television
`camera tube increases with the amount
`of light — from each spot in the scene —
`falling on it. This variable current is used
`to obtain a signal which is processed in
`the studio to yield a broadcast signal.
`The signal
`that
`is
`radiated from the
`broadcasting antenna passes through the
`atmosphere to a receiving antenna where
`it is sensed as a weal: “field strength" —
`a few millivolts per meter of antenna or
`less.
`In the receiver the process is re-
`versed: an increasing positive signal volt-
`age applied to the control grid of the pic-
`ture tube causes an increase in lumi-
`nance at a given point on the screen.
`Thus. in a television camera. an electron-
`
`Introduction
`
`When the average person hears the
`phrase “digital television," he is inclined
`to respond. “I know what 'digital‘ means
`in terms of a digital clock or watch. but
`how can television be digital? Where are
`the numbers?" The answer is that the
`principle of digital television involves the
`use ofnumbers in the generation. manip-
`ulation. recording and transmission of
`television images. but -— unlike the digi-
`tal clock -— the actual numbers are not
`displayed.
`It is the aim of this paper to discuss
`the fundamentals of digital television in
`a simplified way. taking into account the
`widely varying backgrounds of Journal
`readers.
`(Readers well versed in elec-
`tronics or video systems or computers
`may wish to obtain their introduction to
`digital television via the somewhat more
`substantive andchallenging paper by E.
`Stanley Busby,
`.Ir.. which immediately
`follows this one.) We shall in this paper
`compare digital and conventional
`(we
`could say "analog") television and exam-
`ine some of the encouraging possibilities
`and great difficulties facing developers
`of digital television systems. It
`is hoped
`that this paper and the one following will
`provide sufficient
`information for read-
`ers to derive the maximum benefit from
`the more technical and specific papers
`that follow.
`
`Concepts Needed to Understand
`Digital Television
`
`In order to grasp the principles of digi-
`tal television. one should be fairly famil-
`iar with conventional
`television. One
`should also know the significance of cer-
`tain terms which may either be wholly
`new or have special meanings in the field
`of electronics. Such terms include:
`- analog and digital
`538
`
`continuous vs discrete data
`resolution
`nonlinearities and phase distortion
`sampling and quantizing
`modulation techniques
`encoding
`bandwidth
`
`All of these concepts will be covered at
`least briefly in this paper.
`The most direct way to approach these
`mysteries is to clarify by examples the
`distinction between analog and digital.
`There are several analog and digital de-
`vices which by now are familiar to al-
`most everyone. A simple mercury ther-
`mometer and an ordinary slide rule are
`good examples of analog devices; in the
`thermometer. the height of the mercury
`as measured on an appropriate scale is
`proportional — or
`analogous a— to the
`temperature of the surroundings:
`the
`slide rule is a kind of analog computer in
`that
`the distances between the scribed
`lines are made proportional to the loga-
`rithms of numbers so that by adding dis-
`tances one can in effect multiply num-
`bers.
`The ubiquitous electronic pocket cal—
`culator and of course the digital clock
`are obvious examples of digital devices.
`Comparing the slide rule and the pocket
`calculator. we can illustrate two of the
`essential differences between analog and
`digital systems. Analog devices operate
`with continuous data. which means that
`over their operating range any desired
`number can be set in or read out. Digital
`devices. on the other hand. deal with dis-
`crete or stepped data: whatever the least
`significant digit is. it can only change by
`at least one whole unit. And this distinc-
`tion relates directly to the second essen-
`tial difference — that of resolution. Res-
`olution. in the sense used here. provides a
`measure of the smallest possible incre-
`
`JIly 1975
`
`Journal of the SMP‘I'E Volume 8‘
`
`PMC Exhibit 2031
`Apple v. PMC
`IPR2016-00755
`Page 1
`
`

`

`539
`
`ures" - to be able to resolve to one part
`in 256 (0.39%) or better. With this level
`of resolution. the common l-V peak-to-
`pcalt signal used in video processing
`would have each sample rounded off to
`the nearest multiple of 3.9 mV. Experi-
`mental quantizers have been built that
`can resolve at least to one part in 1024.
`(The still photographer often uses the
`principle of digitizing or quantizing
`when he prepares to print a negative with
`an enlarger. He may either make a series
`of test expoaures (say 10. 20. 40 and 80
`seconds long) or he may make one expo-
`sure using a step wedge. Either way, one
`part of the resulting test print is likely to
`be very close to the density and contrast
`that he wants. and he can chooae his fut-
`ure exposures with the enlarger accord-
`ingly. Motion-picture laboratories have
`used step wedges to quantize their print-
`ing exposures in a similar way. Photog-
`raphers,
`furthermore. either minimize
`nonlinearities by hewing to the straight-
`line part of the H & D curve or else use
`them creatively by going to the toe or
`knee.)
`0n the face of it. sampling and quan-
`tizing would seem to work to our disad-
`vantage. Alter all. instead of taking the
`whole voltage waveform as is done in
`conventional
`television. we only take
`pieces of it. Furthermore. there must in-
`evitably be a quantizing or rounding-off
`error made each time we substitute the
`neart quantizing unit step for an actual
`measured value. Nevertheless. if we take
`our samples accurately and frequently
`(the Nyquist Sampling Theorem says
`that the interval between successive sam-
`ples must be equal to or less than one-
`half of the period of the highest frequen-
`cy present in the signal) and then quan-
`tize in small steps to minimize the round-
`ing-off errors, we can use the collected
`and quantized samples
`to recover
`a
`waveform indistinguishable
`from the
`original.
`Still. why is it better to have this col-
`lection of discrete samples
`than the
`whole continuous waveform? The answer
`is that quantized samples can be encoded
`to make up a new signal that in principle
`can be processed. recorded, transmitted
`and ultimately converted back into an
`analog signal
`for playback _ all with
`much less likelihood of errors than the
`original signal could ever be. Instead of
`manipulating the waveform itself —or
`some characteristic of it. such as its in-
`stantaneous amplitude ‘— we can manip-
`ulate information about
`the waveform
`and this information can be used eventu-
`ally to reconstruct the desired waveform.
`
`resent a pattern of light. while in the re-
`ceiver, a pattern of light is generated as
`an "analogy" to represent an electronic
`signal. For color
`transmissions.
`red.
`green and blue spot brightnesses are con-
`verted to currents in the camera and
`back to tiny colored light spots in the re-
`ceiver. The eye of the viewer blends the
`three colors of the spots additiver to
`produce a full-color picture. and since
`the pictures are sent in "frames." persis-
`tenCe of vision permits the viewer to per-
`ceive smooth and continuous motion —
`just as with motion pictures.
`From the small. round. often snowy.
`black-and-white television pictures of
`thirty years ago to the large. sharp. col-
`orful pictures of today.
`the principles
`have scarcely changed. Analog television
`has been greatly refined but it is still an-
`alog. One might compare it further with
`the slide rule: both have been used wide-
`ly for a long time in basically their pres-
`ent form; both require a good eye and
`good judgment to set up and use; both
`have had to contend with problems of
`resolution or reading error: and both are
`now facing competition from digital de-
`vices (pocket calculators and digital sys-
`tems for television studios).
`
`Dealing with the Shortcomings
`of Conventional Television
`
`The word "electronics" did not appear
`in any dictionary printed before I940:
`yet whole new industries — electronic
`data processing and high-speed commu-
`nications (especially television) — have
`grown up since that time. based on elec-
`tronics. The need to process greater and
`greater volumes of data at ever higher
`speeds assured that engineers would
`seize quickly on new technology: electro-
`mechanical
`relays were supplanted by
`vacuum tubes and these in turn gave way
`to transistors and then to integrated cir—
`euits. A number of techniques of data
`processing and signal processing had
`been known in principle for a number of
`years —— but
`their
`actual
`application
`awaited development of electronic sys-
`tems capable of sufficiently rapid switch-
`ing. Thus, digital
`television systems.
`based on some of the same principles as
`the digital voice transmission systems
`used by the military in World War II,
`are only now becoming really feasible.
`When digital
`television was recog-
`nized. just within the last few years. as a
`potentially marketable commodity. there
`was immediate interest in the possibility
`of developing equipment
`to perform
`image processing and control. special ef-
`fects. image storage (as in videotape re-
`cording) and perhaps even transmission
`and reception.
`The obvious place to start was with
`the easily distorted voltage waveform
`produced by the conventional television
`camera. The quality of the picture on the
`viewer’s television set is limited ultimate-
`ly by the purity. accuracy and stability of
`
`this waveform. Yet. there are certain im-
`pairments — degradations of the signal
`and image -—- that tend to be cumulative
`when the signal is relayed many times
`from one link to the next as in tandem
`transmission. Signal nonlinearity. for ex-
`ample. can with each repetition result in
`greater and greater loss of the intermedi-
`ate ranges of contrast. In the film indus-
`try. such nonlinearities are evident in the
`way that spurious dye characteristics
`multiply in duplication stages. Phase dis-
`tortion is a cumulative problem with any
`wideband communications system. but
`especially with high-speed data and
`image transmission systems.
`In televi-
`sion, the effect is of fringing — like dif-
`fraction rings — at edges where the con-
`trast changes abruptly. It is due to une-
`qual delay (phase shifting) of different
`frequency components within the signal
`as they pass through different impedance
`elements— filters,
`amplifiers,
`iono-
`spheric inhomogeneities. etc. The optical
`equivalent
`is seen when white light
`is
`passed through a prism and dispersed
`into colors: red is delayed or refracted
`more than blue.
`Television engineers have worked on
`these and other problems for decades.
`Sometimes one problem is “solved” by
`making another even worse. Even with
`the most up—to—date principles and hard-
`ware, impairments due to phase distor-
`tion and modulation products continue to
`cause difficulty. Now, however. digital
`handling of television signals is opening
`up possibilities that have been difficult
`or impossible with conventional analog
`television systems. These possibilities in-
`clude (1) retiming signals from video-
`tape recorders or from satellite transmis~
`sion systems in which Doppler distor-
`tions have been introduced. and (2) re-
`generating signals at
`intervals along a
`transmission path with the aim of mini-
`mizing the effect of distortions and noise
`introduced by the transmission medium.
`
`Digital Television -—
`Sampling and Quantizing
`In digital television systems. the volt-
`age waveform that is generated by the
`camera to represent the brightness of a
`picture element
`is measured or “sam-
`pled” millions of
`times each second.
`Each sample is then “quantized”: it
`is
`assigned the number of the nearest step
`that the system can resolve. Two kinds of
`quantizing error must be considered here
`because (I) a sample that is exactly be-
`tween two steps can be quantized either
`way and (2) all digits beyond the resolv-
`ing limit will be dropped. {The second
`kind of error is evident on a six-digit cal-
`culator where. for example. the number
`3'}; would be quantized to 3.33333 and
`all subsequent 3's would be dropped.)
`In electronics. such “quantizers” are
`called analog-to-digital (A/D) convert-
`ers. Many of them have enough digits ——
`we could say "places" or “significant fig-
`
`Howell: Digth Television Printer
`
`Modulation and Encoding
`To see why this is so. we must examine
`modulation techniques and encoding.
`There are many forms of communica-
`tion, but virtually all of them impose
`some kind of intelligence on some kind of
`
`PMC Exhibit 2031
`Apple v. PMC
`IPR2016-00755
`Page 2
`
`

`

`Modulatan hove
`
`Currier wove
`
`Amplitude modulotion 1AM)
`
`Frequency mduiollon tFltll
`
`Fig. l. Modulation for continuous-wave or “analog” transmission. A characteristic of one continu-
`ous wave (the carrier) is varied in accordance with another continuous wave (the modulating
`wave). In AM. the varying amplitude (vertical scale) of the modulating signal - proportional. say.
`to the signal put out by a microphone - is impressed on the carrier wave to make it vary in ampli-
`tude. In FM. the modulating signal causes the carrier to vary in frequency while the carrier am-
`plitude is ltept constant.
`
`‘catch‘?" The big problem with digital
`
`"carrier" and this imposing is technically
`called "modulation." Most people know
`of amplitude modulation and frequency
`modulation, but probably not many
`know the significance of these terms or
`the fact that they can be considered as
`examples of analog modulation (see Fig.
`1). Various kinds of discontinuous trans~
`mission (also called pulse-type or dis-
`crete or sampled] are also possible (Fig.
`2). Among these are pulse-amplitude
`modulation (PAM). pulse-duration mod-
`ulation [PDM] and pulwposition modu-
`lation (PPM).
`Under some conditions. it is advanta-
`geous to encode the information repre-
`sented by a signal that has already been
`pulse modulated according to one or an-
`other scheme. For example.
`if a PAM
`signal is encoded. we can obtain a pulse-
`code modulated (PCM) signal. Britain‘s
`Alec H. Reeves invented PCM in 1939
`and found it highly noise-resistant: it is
`therefore well-suited to such communi-
`cations tasks as digital television.‘ The
`encoding in this case would be accom-
`plished by deriving a number proportion-
`al
`to the amplitude of each pulse (see
`Fig. 2c); the number describing the pulse
`amplitude (rather than the pulse ampli-
`tude itself) is then expressed in the form
`of several discrete pulses and is transmit-
`ted in this way. Figure 3 illustrates this
`for the greatly simplified case or the
`noise-resistant
`transmission of a voice
`signal.
`Use of Binary Notation
`electronic
`Because
`the
`simplest
`switches have only two Esaential posi-
`tions. “on” and “off”. it is often very
`convenient to use a binary code to repre-
`sent the sampled amplitude levels. Bina-
`ry codes use only two digits. zero and
`one. and these binary digits
`[called
`“bits“) can be easily represented (as
`shown in Fig. 3) by negative and positive
`pulses. In some systems. nonexistence of
`a pulse is a “zero” and the existence of
`one is a "one"; there are many ways to
`implement a binary code.
`The interpretation of this binary nota-
`tion is straightforward. Where.
`in deci-
`mal notation, each "place" that a digit is
`moved to the left multiplies the value of
`that digit by ten. in binary notation. each
`place moved multiplies the value of the
`digit by two. Thus "ID" in binary is “2"
`in decimal: “lfll” in binary is "5" in dec-
`imal; and “l 1 1011i“ in binary (Sample
`6 in Fig. 3) is l 19 in decimal.
`The “places” available in a given bina-
`ry number code limit both the largest bi-
`nary number that can be expressed and
`the resolution that can be obtained in
`that binary code. The largest binary
`number that can be expressed with n bits
`is 2" - 1. and the resolution is limited to
`* Phase-modulated PCM is also popular and might
`be considered to be derived From pulse-position mod-
`ulation. but
`in the present discunsioll we shall deal
`only with the form of FCM that in essence is encod-
`ed PAM.
`
`one part in 2". Thus. four bits give a res-
`olution of one part in 16. five hits one in
`32, six bits one in 64. seven bits one in
`[28 and eight bits one in 256. (For com-
`parison, the highest three-place decimal
`number is 999 and this permits a resolu-
`tion of one in I.000— since zero is also
`a possibility.)
`It appears that eight-bit
`resolution (where eight bits “describe”
`each sample) may be the minimum that
`is acceptable for broadcast television.
`Pulse Code Modulation and
`Correlation Techniques
`The effectiveness of pulse code modu-
`lation is well illustrated when it is used
`with computer correlation techniques -—
`although no one has suggested that such
`techniques be used with television sys-
`tems. If pulse code modulation of the
`kind shown in Fig. 3 is used. we know a
`priori that the signal is either a positive
`pulse or a negative one. Computer corre-
`lation techniques can then be used at the
`point of reception to determine if the sig-
`nal correlates with positive or with nega-
`live. In correlation detection. a signal is
`compared point to point with an inter-
`nally generated reference. The output of
`such a detector is a measure of how
`closely the input signal
`resembles the
`reference. The reference signal is at all
`times a “best guess” by the computer of
`what the input signal should be at that
`time. The noise (which can be much
`larger than the signal). having a random
`nature, will be uncorrelated and there-
`fore eliminated when correlation tech-
`niques are employed.
`Pulse-code modulation and computer
`correlation detection have been used very
`effectively in aerospace applications such
`
`tel
`
`Fig. 2. Pulse modulation techniques available
`to the communications engineer. (a) The mt»
`dulating signal; (b) the carrier: (c) pulse am-
`plitude modulation (PAM); (d) pulse dura-
`tion modulation (PDM) - also called pulse
`width and pulse length (PWM and PLM); (e)
`pulse position modulation (PPM).
`
`as constructing radar maps of the planet
`Venus. Although the signal sent out has
`rnany kilowatts of power. the return sig-
`nal is only on the order ofa quadrillionth
`of a watt and deeply embedded in noise.
`Nevertheless. computer correlation de-
`tection of s PCM signal can not only
`show that the signal is there but also de-
`rive useful information from it.
`
`Limitations and Prospects
`of Digital Television
`Everything we have said about digital
`television so far has been positive. The
`alert
`reader. however. should be just
`about
`now asking.
`"What
`is
`the
`
`540
`
`July 1975
`
`Journal of the SMP’I'E Volume 84
`
`PMC Exhibit 2031
`Apple v. PMC
`IPR2016-00755
`Page 3
`
`

`

`ORIGINAL SIGNAL
`
`I28tEVELS
`
`l23456
`
`3.000 SIMPLES PER sacouo——r
`
`Sample I I 65
`
`Salling 2 ‘ 34
`
`Sample 3'76
`
`Sample 4- HO
`
`50!!! 9|. 5 ' 95
`
`Sell-tell s-ll9
`
`0
`
`O
`
`the idea of transmitting digitized televi-
`sion pictures from satellites to ground
`stations but only in rt tight beam and
`only with six-bit resolution.) Still.
`in-
`tense efforts are being made to circum-
`vent— to any degree possible and by
`any means possible—the requirement
`for such a large bandwidth.
`In the face of such difficulties. how
`can communications engineers be san-
`guine about the future of digital televi-
`sion sySIerns'! They can because the
`technology is evolving. More efficient
`encoding techniques are being developed
`continually. The use of lasers and light
`guides can. in principle. permit the trans-
`mission of thousands of television pro-
`grams simultaneously over a single fiber.
`Integrated circuit (10 technology is ad-
`vancing at a fast pace. and the cost of IC
`units is coming down.
`Finally, the motivation to develop dig-
`ital television certainly exists — and this
`does not just mean to get a clearer pic-
`ture on the screen of the viewer at home.
`The biggest motivating factor may be to
`get digitized signal processing equipment
`into studios. With such equipment. spe-
`cial effects.
`image enhancement and
`many other functions could be handled
`with great case;
`it
`is likely that many
`could even be automated.
`It is certainly going to be interesting
`to watch the development of digital tele-
`vision over the next decade or so. It is
`hoped that this paper will aid the reader
`in understanding some of the develop-
`ments as they occur.
`
`The BKSTE Jam. 57: l Ill-115. April IQTS.
`
`Fig. 3. When PCM is used to tt-artstttit a voice signal. the amplitude of the original were is sampled
`8.000 times per second and the sampled values are translated into binary code groups consisting of
`1‘s and 0’s (positive and neth pulses). Code groups seven binary digits long malt: it possible to
`measure the instantaneous amplitude of the original wave to an accuracy of one part in 128.
`Often. an extra digit is added to each group of seven for signalling and other functions. making
`the bit rate 64.000 per second (8.000 K 8). At the receiver, the sequence of pulses is decoded to
`obtain the original signal.
`
`television. at present. can be summed up
`in one word: bandwidth. Bandwidth is
`the difference (often measured in mega-
`hertz) between the upper and lower lim-
`its of a frequency band. The Nyquist
`Sampling Theorem and the need for at
`least eight-bit binary numbers to achieve
`adequate resolution conspire against us
`to aggravate the wide bandwidth prob-
`lem that is already the preeminent curse
`of television. Because an ordinary color
`television signal has a bandwidth of
`about 4.5 MHz and every individual
`cycle can be important. we must sample
`approximately ll million times each sec-
`ond (as opposed to the "mere" 8.000
`times per second in the voice example
`considered earlier). Using an eight-bit
`binary number to describe each sample.
`it is obvious that we would need to be
`able to process about 88 million bits per
`second.
`
`To drive home the sheer magnitude of
`this number. consider that transmitting
`this quantity of information is equivalent
`to sending the Bible (Old and New Tes-
`taments) more than ten times in the
`course of a second. And that is with the
`most efficient encoding theoretically at-
`tainable. To transmit
`information at
`such high rates. one needs an informa-
`tion upipeline“ with a greater "diame-
`ter“ than anything so for used for televi-
`sion. And it is more than just a technical
`problem: “pipelines” take up “real es-
`tate“; the airwaves are not unlimited and
`other users have their rights too.
`This extraordinary bandwidth prob-
`lem is the primary reason that no one at
`present is even considering broadcasting
`this data stream. but only using it be-
`tween thc analog camera and the analog
`transmitter. (Members of the European
`Broadcasting Union are working with
`
`Bibliography
`l. .l. R. Pierce. Symbols. Signals and Noise. Har-
`per Jr. Row. New York. I96].
`1. Henri Busignics. “Communications channels."
`Scientific American. 227: 99-] I3. Sapt. [972.
`3. W. M. Goodall. "Television by pulse code mod-
`ulation." Belt Sysr. Tet-It. Jam. 30: 33-49. Jan.
`l9SI.
`. A. A. Goldberg. "PCM-encoded NTSC color
`television subjective tens." Jour. SMPTE. 82:
`«toast. Aug. I913.
`. Woodrow W. Everett. Jr.. Topin .t'n Interryste'm
`Electromagnetic Compatibility. Holt. Rinehart
`and Winston. New York. 1912.
`. Clyde F. Combs. Jun. Ed. Basie Electronic ht-
`.rtrtmmtt Handbwlc. McGraw-Hill Book Co.
`New York. [972.
`. Ray Ryan. Basic Digital Electronics. Tab
`Books. Blue Ridge Summit. Pa.. 1915.
`. Albert Race. Vision — Human and Electronics.
`Plenum Press. New York. 1973.
`. Pierre Merta.
`“Long-haul
`television signal
`transmission." Jot-Ir. SMPTE,
`3'5: 350-355.
`Sept. I966.
`. C. W. B. Reis. “Tutorial — Anyone ior digits?"
`
`Howell: Digital Telerlsiott Printer
`
`PMC Exhibit 2031
`Apple v. PMC
`IPR2016-00755
`Page 4
`
`