HE INFORMATION presented in this section has been compiled from several modern and historical sources and, errors and omissions excepted, the intention is to give a summary of the various standards at the time that they were current. Nevertheless, it is hoped that present-day standards are also accurately accounted for, and to this end any corrections would gratefully received (please E-mail me with any comments). Thanks are especially due to Mark Carver, Steve Palmano and Peter Vince for help and advice. Written sources consulted include [Electronics and] Wireless World and [Practical] Television magazines, textbooks by Benson KB and Whitaker JC, Carnt PS and Townsend GB, Holm WA, Hutson GH, Kerkhov F and Werner W, and technical publications from BBC, EBU, IBA and ITU.

I am particularly indebted to Peter Vince for recently spotting certain anomolies in the ITU document BT.470-6 from which many of the details in these pages were taken. It has been superseded by BT.1700 and BT.1701, and the values quoted in these pages are now verified by those, and by SMPTE 170M-1999 in relation to the NTSC standard. Many of the NTSC parameters feature recurring decimal fractions, and I have indicated these throughout with square brackets, for example f_SC = 3 579 545.[45]Hz

OME SIGNIFICANT dates in the comings and goings of television line standards, colour standards and transmission bands. Starts and ends of official services are in roman type, while other landmarks and experimental services are in italics.

Oscilloscope trace of a line of video containing a staircase pattern (screen shot inset)

LL ANALOGUE television systems work in the same way - only the precise characteristics are different. The image is dissected in much the same way that you are reading this text. A sampling device scans across from left to right, and from top to bottom, in a series of near-horizontal lines that are arranged in a rectangle to form a raster. The output from the sampling device comprises a constantly changing voltage which at any moment represents the brightness of a given point in the image.

This video stream is punctuated at the start of each scanning line by a line synchronising pulse, and at the start of each frame, or field, by a series of field synchronising pulses. The sync pulses are separated from the picture information in both time and amplitude. In a standard 1Vp-p (into 75 ohms) 'composite' (mixed syncs plus picture) video signal, the peak white amplitude is +700mV, whilst black and blanking levels sit at 0V and the sync tips reach -300mV. In the 525-line standard however, peak white is 714mV and sync level is -286mV. In addition, in the US (but not the Japanese) version black level is 54mV above blanking level.

The most basic difference between television standards is the number of lines per field and the number of fields per second, as determined by the line and field scanning frequencies. Details may be found in the line standards section of this page.

HUNDRED years ago or more, the moving picture industry satisfied itself that the persistence of vision effect in the human eye and brain was such that by projecting a series of still images at the rate of sixteen or eighteen per second, an illusion of smooth movement was created. A few year later, with the introduction of synchronised sound, the projection rate was increased, and standardised at twenty-four pictures per second, partly to give smoother lip movement, and partly to increase the writing speed available for the optical sound-track. However, sixteen, or even twenty-four, complete blackouts per second between pictures creates too much flicker for the human brain to tolerate. For that reason a shutter is placed in the light beam of the projector that interrupts it forty-eight or seventy-two times a second, giving a flicker-free impression of smooth movement.

It would be a simple matter to incorporate such a scheme into a modern television system. All it would require is that each frame, as it is received, is written into a digital field store (an integral part of every digital tv receiver) and read out again two, three, or even four times at a much faster rate.

Unfortunately, digital field stores were not available in the nineteen-thirties when analogue television was being developed. Instead, having determined that at least 24 complete frames per second, each with at least 240 scanning lines, were required to provide a watchable picture, the designers came up with an ingenious method of reducing the flicker rate. Using an odd number of lines per picture they simply doubled the field frequency, whilst keeping the line frequency the same. So, in the 405-line standard, the scanning beam reads 202.5 lines over the complete height of the image, then half-way through line 203 it jumps back to the top of the image to read the 202.5 lines that lie in between the first set.

In this way there is still a full 405 lines per picture (though in practice, because the video signal is time-shared between picture information and synchronisation pulses, there are actually only 377 lines of picture per frame, not 405), but there are now fifty fields per second projected onto the tv screen to reduce flicker. As a bonus, because of the way the television camera works, each picture line contains information from the last 1/25 second period since it was last scanned, and 1/50 second earlier than that for the lines immediately above and below, and so this 2:1 interlaced system of alternate staggered fields has the effect of smoothing out movement even further.

Ironically, now that receivers with digital field stores are with us and it is possible to increase the displayed picture repetition rate beyond fifty per second, the presence of 2:1 interlace causes huge problems for tv set designers and creates nasty motion artefacts on the screen, which are diffcult to eliminate.

All 'standard definition' (ie between around 400 and 900 lines per picture) analogue television systems incorporate 2:1 interlace.

HE ASPECT ratio of a picture is its width divided by it height and is often expressed as the ratio of two integers. The original aspect ratio of the 405-line standard was 5:4, but it was later changed to 4:3 to be the same as the so-called 'Academy Ratio' of 35mm cinema films. More recently, with the advent of digital transmissions, a second change has been made, to 16:9, and this 'widescreen' format is running side-by-side with 4:3 in many countries. Both 4:3 and 16:9 pictures use exactly the same portion of the transmitted signal. The latter format is sometimes called 'anamorphic' by analogy with the cinema format in which a cylindrical lens is used to squash a wide picture 'anamorphically' into a standard Academy Ratio frame at the expense of reduced horizontal resolution.

Other aspect ratios may be created within 4:3 or 16:9 frames by effectively widening the horizontal or vertical blanking periods in order to matte down the visible picture to the required shape. British broadcasters frequently use this technique to present a 14:9 version of a 16:9 picture on analogue transmissions. Since these extra pseudo-blanking periods are really part of the active picture it is possible to include graphics in them. In particular sports and light entertainment producers like to add swirling coloured 'curtains' to 4:3 or 14:9 segments of their widescreen shows.

There is more about aspect ratios in 'More than just a pretty face...'

ESOLUTION, OR 'definition' is a measure of the fineness of detail that can be seen in a picture. In photography, the resolution is generally the same horizontally and vertically, but in television the two are separate, though interdependent. Vertical resolution is determined by the number of scanning lines in the picture, and horizontal resolution by the video bandwidth available. In most line standards the two are made to appear equal to the eye, though there is some disagreement about what constitutes equality. One solution has been to apply a 'Kell factor' in calculations to determine resolution.

There is more about resolution on the page 'More than just a pretty face...' in the Test Cards section of this web site.

In colour television systems the chrominance (hue and saturation) resolution is generally much lower than that of the luminance (black and white information). This is discussed in the chapter on Colour below.

ISPLAY TUBES are not linear devices (though camera pick-up tubes tend to be). Typical cathode ray tubes produce a light output that is proportional to the driving voltage raised to the power of 2.8 ±0.3, known as the 'transfer characteristic' or 'gamma factor'. This is due mainly to the triode transfer characteristics of the display tube.

To correct for this non-linearity, the video signal should require an exponent of 1/2.8 or 0.357, but it has been found that an overall system gamma of unity renders monochrome pictures which appear flat and lacking in contrast. A value for gamma correction of 0.45, giving an overall system gamma of 1.26, has been chosen in non-NTSC countries for monochrome transmissions.

However, the various equations used to matrix the colour signals require an overall system gamma of unity in order to yield correct colorimetry, so when using these equations a precise transfer characteristic of 1/0.4545 = 2.2 is assumed, and a gamma correction value 0.4545 is appled to all colour standards, despite the fact that a display transfer characteristic of 2.8 is still assumed for 625-line colour standards.

Values of 2.2 for transfer characteristic and 0.4545 for gamma correction are used in the NTSC countries, leading to a system gamma of unity. For computer displays, Windows assumes a transfer characteristic of 2.2 while Macintosh uses 1.8, leading to a low overall gamma of 0.82, in which low luminance levels are rendered brighter.

It has always been the practice to perform this gamma correction in the camera (or the colour encoder) in order to reduce the complexity of receiver video circuitry, and to reduce the effects on dark parts of the picture of noise accumulated in the transmission system.

Signals that have been gamma corrected should properly be written with a prime mark ('), for example Y', R'G'B' (or, when referring to the voltages: E'_Y, E'_RE'_GE'_B). However, since most signals are gamma-corrected I have left out the prime marks in general to avoid cluttering up the text. They are included in some of the equations in order to clarify which values are gamma-corrected and which are not.

AVING ESTABLISHED workable monochrome television systems the designers turned their minds to colour.

Original colour scene

RGB

To reproduce a colour scene requires the image to be sampled separately in the three additive primary colours red, green and blue (R, G and B). In colour photography, printing and computing, it is usually these three colour separations, or their subtractive counterparts (C - cyan, M - magenta, Y - yellow and K - black, or key), which are stored, manipulated and displayed. However, the legacy of the monochrome transmitters and receivers all over the world, together with the huge amount of frequency spectrum that would have been required, meant that a different approach was needed for colour television.

At this stage of the process the levels are normalised such that for peak white, R = G = B = 1 ^[1], and gamma correction is applied to the three colour separation signals as it is these that will be used to drive the cathodes of the three colour display tube guns.

^[1] See the section on Component video levels below for the actual voltages used.

Colour separation signals

R	G	B

It was recognised that any colour system should be compatible in both directions - ie no change should have to be made to monochrome receivers, and colour sets should display monochrome transmissions correctly and automatically. The black-and-white picture was therefore redefined as 'luminance' (Y - not to be confused with the yellow component of CMYK colour space) and is synthesised by adding together the three separate colour separation signals in the proportions Y = 0.299R + 0.587G + 0.114B, these values having been determined to produce a compatible 'panchromatic' display on a monochrome receiver. Again, for peak white, where R = G = B = 1, Y = 0.299 + 0.587 + 0.114 = 1.

Luminance signal

Y

This luminance signal is transmitted in exactly the same way as the old black-and-white signal.

Now that some of the colour information is effectively coded in the luminance signal, it is only necessary to transmit two further signals in order to be able to obtain the separate R, G and B signals in the receiver. The method that has been universally adopted is to matrix the R and B signals with the Y signal and transmit (R-Y) and (B-Y), where (R-Y) = 0.701R - 0.587G - 0.114B, and (B-Y) = - 0.299R - 0.587G + 0.886B. This has the huge advantage that in the case of a monochrome picture, or areas of grey in a colour picture, the colour values are such that Y = R = G = B and therefore (R-Y) = (B-Y) = (G-Y) = 0. In other words these 'colour difference' signals vanish when there is no colour information, improving compatibility and reverse compatibility (since a colour receiver, seeing a monochrome transmission with no colour-difference signal present, will automatically display a black-and-white picture).

Colour difference signals

(R-Y)	(G-Y)	(B-Y)

Being 'difference' signals, the (R-Y), (B-Y) and (G-Y) voltages, unlike other video signals, can be negative as well as positive. For the purpose of these illustrations I have added a mid-grey pedestal so that the 'negative' excursions are visible. The grey areas represent zero colour difference voltage (ie colourless areas of the picture) and because the eye is sensitive to very small changes in the 'colour temperature' of neutral shades, the three signals have to be very accurately clamped to 0V in the coder and decoder circuitry.

The (R-Y) and (B-Y) signals were chosen for transmission because they have larger maximum voltage excursions than the (G-Y) signal, which is therefore recoverable in the receiver by attenuating, rather than amplifying, the other two. This has advantages in terms of signal-to-noise ratio as well as decoder complexity. The green colour difference signal is given by (G-Y) = - 0.509(R-Y) - 0.194(B-Y), and in early colour receivers the addition of the Y signal ('matrixing') to recreate the R, G and B signals, was done within the crt display tube, thus economising further on valves.

Colour difference displays

(R-Y)	(G-Y)	(B-Y)	(R-Y), (G-Y), (B-Y)

The four pictures above show the displays obtained when the colour difference signals are applied individually and in combination, in the absence of the luminance signal, to the matrix circuitry from which the final R, G and B signals are extracted. When the luminance signal is also applied to the matrix the original colour separation signals are obtained as shown below.

Colour separation displays

R	G	B	RGB

These final four pictures show the displays obtained when the colour separation signals are applied individually and in combination to the display device.

The above process is essentially the same for all analogue and digital colour television systems.

The next problem was to accommodate this three-fold increase in information without increasing the bandwidth of the transmitted signal. Two phenomena - one physical and one physiological - allowed this to happen.

Colour vision acuity, it was discovered, is different from that of brightness-only vision. In fact, if sufficient detail is available in the brightness of a scene, the detail in the colours can be reduced considerably with no apparent reduction in the sharpness of the scene. This allows the bandwidth of the colour difference signals to be reduced to half, or less, of that required for the luminance channel. This fact is further exploited in the NTSC system, where the colours to which the eye is least sensitive in terms of detail are assigned a narrower bandwidth, by transmitting I (orange-cyan) and Q (green-magenta) signals instead of (R-Y) and (B-Y) and allowing the Q signal only half the bandwidth of the I signal.

Secondly, the video waveform does not have a continuous frequency spectrum, like that of sound. Because of the way the picture is scanned, the energy in the spectrum is bunched around multiples of the line and field scanning frequencies, with little energy in the gaps. By shifting the colour-difference signals in frequency (by modulating them onto a subcarrier) it is possible to make the peaks in colour energy fall in the gaps in the luminance energy, thus allowing the signals to be separated at the receiver by means of a 'comb filter'. The subcarrier frequency must be high enough that the dot-matrix pattern created on a black-and-white receiver is not too coarse, yet must be low enough that the upper chrominance sideband fits within the vision bandwidth of the transmitted signal without attenuation or distortion. The precise value of the subcarrier frequency is then determined by adding fractions of the line and field scanning frequencies in order to create a dot pattern that is not distracting.

Thus the three channels of colour information may be fitted into the same bandwidth as existing black-and-white video signals, whilst maintaining both forward and reverse compatibility. The differences between the three main colour systems occur in the way the colour difference signals are modulated on the subcarrier, and the precise frequency of the subcarrier (which depends to a large extent on the line and field frequencies and the bandwidth of the transmitted signal). These are detailed in the colour standards section of this page.

Causes of loss of resolution in the decoded colour signal

Although for most picture content the reduction in chrominance bandwidth is perfectly acceptable, in certain circumstances it can lead to unwanted visible effects, especially at the boundaries between saturated colours or where brightness detail occurs in areas of saturated colour. Captions, for example, in red or blue appear smeared and fuzzy. Also scenes illuminated by light of a primary colour appear noisy and blurred.

Here is a waveform of a multiburst test signal. It ranges from black (0%) to white (100%) with a mean level of 50%. In a digital transmission all the gratings up to 6MHz will be visible on the screen. In an analogue 625-line signal the 6MHz grating will appear plain grey, as will the 5MHz grating in a 525-line signal, because they are beyond the frequency response of the luminance channel. However, the response of the chrominance channels is much less - around 1MHz for PAL, SECAM and NTSC, and around 3MHz for digital.

Let us see what happens if we transmit a multiburst entirely in one primary colour. The diagrams that follow show one grating that is within the chrominance passband followed by one that is outside it. The signal is applied to the red channel only, so the green and blue channels are at black level (0%).

The luminance and the three colour-difference channels are shown here. Adding together the luminance and each of the colour difference signals in turn will give the same R, G and B signals as above.

However, the chrominance channels are low-pass filtered. The lower frequency grating remains unchanged, but the higher one becomes a straight line having the same mean level as the grating.

When the luminance signal is added to each colour-difference signal, the amplitude of the high-frequency grating is reduced in the active channel, and phantom gratings appear in the other two channels, reducing the saturation (and fortuitously increasing the brightness slightly) of the details. The red and green channels contain negative-going information which will either be clipped by the display circuitry or cut off in the cathode ray tube.

The loss of luminance resolution is less pronounced in areas of saturated green and more pronounced in saturated blue because of the relative contributions made by the colour separation signals to the luminance signal.

This sequence of screenshots shows the two sections of multiburst as a black-and-white signal...

..as a signal in the red channel only as displayed on an RGB monitor before encoding...

..and as displayed after decoding.

This effect is increased by the way that gamma correction is applied in generating the luminance signal. As discussed above, the luminance signal is derived from the three colour separation signals using the following relationship:
Y = 0.299R + 0.587G + 0.114B
in which the uncorrected luminance signal is obtained by summing the uncorrected colour separation signals. The gamma corrected luminance signal Y' is then obtained by applying gamma correction to Y.

However, in terms of practical circuitry, it is less complicated to derive Y' by summing R', B' and B' as follows:
Y' = 0.299R' + 0.587G' + 0.114B'
which unfortunately has the result that more of the high luminance frequencies are transferred from the Y' signal to the (R'-Y') and (B'-Y') signals after matrixing. These are then filtered out by the coder and lost, resulting in yet more reduction of the resolution.

An additional effect is produced because many television cameras and telecine units employ a technique called 'contours out of green' in which the the waveforms used for horizontal and vertical aperture correction are derived not from the luminance signal obtained by matrixing the three colour channels, but from the green channel alone. Aperture correction is a technique used to compensate for the fact that when scanning is employed, the sampling device - the electron beam in a camera tube, the spot of light in flying-spot telecine or the 'pixel' in a charge coupled device - is of finite size, which limits the resolution of the device in a predictable and correctable manner. The addition of a 'crispening' signal to the three colour separation signals can overcome this deficiency, and deriving it from the green channel alone improves the sharpness of the picture, since fine detail in all three channels might not be perfectly coincident due to poor registration.

Since most scenes contain high levels of green, there are usually no unwanted effects, but when the pictures contain large amounts of saturated red or blue detail, as in the case of red captions on a film, or 'disco' scenes, the pictures appear blurred because the green channel is contributing little to the aperture correction circuitry - other than a high level of undesired noise - and so the colour separation signals contain only the 'unsharpened' video information. Unlike the effect described above, this effect is also seen on monochrome receivers, where pictures with low green content lack high-frequency definition. Not all cameras use 'contours out of green'. Indeed Sony developed one model in the late 1970s that used 'contours out of red' specifically for photographing surgical operations.

A further problem is that with scenes illuminated by a primary colour limiting can occur in the channel of that colour, resulting in no detail at all being visible in the clipped areas. Again, this also affects monochrome displays and leads to odd looking pictures because the clipping occurs at low luminance values rather than peak white.

Developments in technology, especially the area of home video recording, have led to improved, though non-compatible (with old B&W receivers), ways of delivering colour signals at baseband rather than radio frequency. Where the standard coded colour signal ('CVBS' - Colour, Video, Blanking, Syncs) carries luminance and chrominance within a single circuit 'S-Video' ('Separate video' - not to be confused with the tape format 'SVHS' - Super VHS, where recorders often incorporate S-Video inputs and outputs) carries the luminance in one circuit and the coded chrominance in a second, improving the bandwidth available to each whilst eliminating crosstalk. With digital systems the luminance and two colour difference signals are encoded separately, and so digital decoders can be made to generate RGB, YP_bP_r (a version of Y(B-Y)(R-Y) - see Standard Video Levels below), S-Video or CVBS depending on the capabilities of the display device.

The colorimetry - that is the precise colours used for the primaries red, green and blue, and also 'white' - differs from standard to standard, and has also changed over the years. These are described and discussed in the colour standards section of this page.

Finally, the word 'COLOUR' may appear real enough in the caption on the left, but when viewed on a greyscale monitor it will disappear, since it has been arranged that all the information pertaining to it appears in the colour difference channels and none in the luminance channel.

E HAVE already seen that the standard monochrome composite video signal has a peak to peak amplitude of 1V into 75ohm. With blanking level at 0V, sync tips are -300mV and peak white is +700mV. This is the same for the luminance component of a colour signal as well, though the levels of the colour components depend upon the use to which the signal is to be put.

100% colour bars (for further details of colour bars and other test patterns see the Colour Bars section of the Test Cards page) provide a source of the extremes of voltage excursions allowed by the colour tv system (its 'gamut'). In the diagrams below, the colour of the trace indicates the signal being considered and the background colour indicates the bar being displayed. Note that the waveforms below represent actual voltages, while the values indicated in the Colour Bars section are normalised - that is, black = 0 and white = 1.

R	G composite	B

The gamma corrected non-composite red, green and blue colour separation signals each have a maximum amplitude of 700mV. Synchronisation is either by separate wire(s), carrying mixed syncs or separate H and V sync pulses, or by incorporating mixed syncs onto one (usually green) or all of the separation signals. The domestic SCART system uses non-composite 700mV RGB signals with the display being synchronised by the accompanying encoded composite signal, which has to be carried for compatibility with non-RGB input capable displays.

Y composite	B-Y	R-Y

These values of RGB yield luminance (Y) and colour difference (B-Y and R-Y) signals of amplitudes 0-700mV, ±620mV and ±491mV respectively. These are the (relative) values that must be matrixed to recover the correct RGB signals in the decoder, but the colour difference signals are unsuitable for distribution at these levels.

PAL composite	U	V

To produce an encoded PAL signal the colour difference signals, after scaling, are amplitude modulated onto two supressed quadrature subcarriers. In order to achieve levels of subcarrier that can be further amplitude modulated onto a vision carrier, whilst maintaining reasonable signal-to-noise ratios for the two signals, the weighting factors are as follows:

E'_U = 0.493(E'_B - E' _Y)
E'_V = 0.877(E'_R - E' _Y)

These particular weighting factors ensure that the maximum subcarrier excursions are around 33% above white level for saturated yellow and cyan colour bars and 33% below black level for red and blue bars.

The p-p subcarrier amplitude (indicated on the composite waveform by blocks of saturated colour) is twice the vector sum of the amplitudes of the U and V signals for each bar. However, because the frequency response of the system is not always perfect, the recovered subcarrier amplitude, and hence the amplitudes of the demodulated colour difference signals, may be higher or lower than this. In the PAL and NTSC standards that would affect the saturation of displayed colours, so the colour burst signal (shown here in grey) inserted into the front porch of the line blanking period in order to synchronise the reinserted local subcarrier has an amplitude of 300mV p-p which is used as the reference level for the automatic chroma gain control to ensure that the subcarrier is demodulated with the correct amplitude.

The three purple plimsoll lines at the top left of the composite waveform represent zero carrier level in Systems I, B/G/D/K and M, for which the modulation depths for peak white are respectively 20%, 15% and 10% (100% modulation is sync tip level in all three cases). It is clear that none of the systems can safely carry the full gamut of 100% saturated colours without severely distorting the transmitted signal.

Y composite	Pb	Pr

Although four-wire (with additional control wires) RGB signals carried by the French SCART (Peritel) interconnection system have been in use in Europe since the early 1980s, the preferred method in NTSC countries is three-wire component, called YP_bP_r. The colour difference signals are individually scaled in order to obtain a maximum amplitude of ±350mV for them both, the same swing as the black-to-white portion of the luminance signal. The scaling factors are as follows:

E'_Pb = 0.564(E'_B - E' _Y)
E'_Pr = 0.713(E'_R - E' _Y)

Y composite	Cb	Cr

When component signals are digitised a 350mV pedestal is added to the scaled colour difference signals in order to bring them into the same 0-700mV range as the luminance signal. The trio is then termed YC_bC_r. The luminance signals are sampled at 13.5MHz and the chrominance signals at 6.75MHz for both 525- and 625-line standards, as specified in the document ITU-R BT.601. The eight-bit quantisation levels (range 0-255 decimal) corresponding to the 0, 700mV luminance signal are 016 and 235 decimal. For the 0, 350, 700mV colour difference signals they are 016, 128 and 240 decimal. The negative-going synchronisation pulses are not digitised since they are outside the sampling period (720 samples @ 13.5MHz over 576 scanning lines in the 625-line system and 480 lines in the 525-line system).

HE FINAL leg of the journey is to get the video, and associated audio signals, into the home. The art of radio transmission was quite mature at the time of the introduction of television, but to transmit the vast bandwidth of a video signal (one thousand times wider than an audio signal) required carrier frequencies of an order of magnitude higher than those that had been used previously for broadcast purposes.

Also, the theory of modulating a varying signal onto a carrier wave had to be rethought. Because audio signals are symmetrical about a mean value, it is only necessary to ensure that that value equates to 50% of the carrier level of the output of the transmitter. As the modulation level increases, the positive and negative peaks move towards 100% and 0% carrier respectively.

However, a vision signal is not like that. The negative part of the signal between 0 and -300mV is constant, and carries the synchronisation pulses. The positive portion between 0 and +700mV carries the vision proper, and depending on the amount of white in the picture the mean level of the signal can vary between around -30mV and +950mV. Such a signal would be useless for a modulator that expected a constant 50% mean level, because the positive and negative peaks could float above 100% and/or below 0% carrier levels according to picture content. Alan Blumlein introduced the concept of DC restoration, whereby the video signal presented to the modulator is 'clamped' so that blanking level represents a fixed modulation level.

Oscilloscope trace of a System L carrier modulated with two lines of monochrome video using positive modulation

Oscilloscope trace of a System I carrier modulated with two lines of monochrome video using negative modulation

The first frequencies chosen for television transmission were around 50MHz in the hitherto unexplored vhf band I (Alexandra Palace had vision at 45.0MHz and sound at 41.5MHz). Amplitude modulation was used for the vision signal and although the original Alexandra Palace transmitter radiated the full double-sideband signal, all later transmissions have been vestigial sideband, with one of the sidebands filtered out beyond the first few hundred kilohertz. The sound carrier was placed just beyond the radiated vision sideband, where the video modulation energy was relatively low.

The possible options of vision modulation sense, sound modulation mode and vision sideband supression, in addition to the various line standards, has led to a multitude of different transmission standards around the world.

Vision modulation may be either positive or negative. With positive modulation, the sync pulse tips are held at the zero-modulation level, whilst peak white is 100% and black level around 30%. With negative modulation, the sync tips are at 100%, black and blanking levels around 75% and peak white 10-20%, depending on the precise transmission system used. This method has the advantage that there is a portion of the waveform that is always at 100% modulation, so that the receiver can measure the carrier strength and adjust its automatic gain control accordingly.

The sound carrier can be amplitude or frequency modulated. The convention is that amplitude modulated sound is used with positive vision modulation because in the intercarrier sound method of frequency modulation detection the sound carrier and vision carriers are mixed together to give an accurate intermediate frequency set by the transmitter, allowing them to share a common intermediate frequency amplifier chain in the receiver. In positive vision modulation the carrier level falls to zero during sync tips, making it unsuitable for this purpose and so AM sound, which is not so sensitive to local oscillator drift in the receiver, is used, with a separate intermediate frequency amplifier from that used for the vision signal. However, AM sound requires a larger amplitude of carrier than FM for the sound and vision service areas to match.

The choice of which vision sideband to supress is immaterial for most purposes, except that it affects the position of the sound carrier relative to the vision carrier. Indeed, the French System E 819-line network had a mixture of upper and lower sideband transmissions shoehorned into bands I and III in order to provide more useable channels.

Analogue direct-to-home satellite broadcasts by comparison use frequency modulated video with a vision bandwidth of just over 5MHz and several frequency modulated sound carriers (used in pairs for stereo) between just below 6MHz and 8MHz. The channel width is about 27MHz compared with the 8MHz of terrrestrial systems B,G, I D and K.

Stereo and multilingual soundtracks have been added to analogue terrestrial transmissions in many countries. These have been incorporated either by multiplexing the existing sound carrier or by adding further analogue or digital carriers. These are detailed in the table of CCIR transmission systems.

HEN A carrier wave is amplitude modulated, its amplitude varies in sympathy with the modulating waveform. This is shown for a video transmission in the diagrams above, where the envelope of the waveform is shown as a full line. In reality the envelope is simply defined by the peak tips of the carrier wave and the modulating waveform itself is not sent. These diagrams show what is happening in the time domain.

Frequency spectrum of an amplitude modulated telephony signal The frequencies marked are relative to the carrier. The levels and slopes of the curves are stylised for clarity.

It is not obvious what happens in the frequency domain. In fact when two sinusoidal frequencies (pure tones) are mixed together in a non-linear way (as happens in amplitude modulation) the result comprises the two original frequencies as well as their sum and their difference. In a radio signal, one of these frequencies is called the carrier - this is the frequency to which you tune your receiver. The original baseband modulating frequencies are filtered out of the transmission as they are not required. In an audio am transmission for example, the rf portion that is transmitted comprises the carrier wave itself, plus the sum of it and every audio frequency in the modulating signal gathered together in what is called a sideband. A mirror image of this sideband comprises the difference between the carrier frequency and all the contributing audio frequencies. These are called the upper and lower sidebands.

Note that the amplitude of each sideband is half that of the carrier, and that the bandwidth of the transmission is twice that of the highest modulating audio signal, which in this case is around 3kHz (it is a telephony communications, rather than entertainment broadcast, signal). In the simplest form of demodulator, called an envelope detector these three signals are used to recover the original audio signal, which was filtered out of the transmission.

Now, the information contained in each sideband is exactly the same, and to send it twice is wasteful of bandwidth and power. The carrier wave, once it has been used to generate the two sidebands, carries no information whatsoever - either sideband would fly just as far without it. In telephony communications therefore, both the carrier wave and one sideband are often filtered out, and the receiver regenerates the original audio - or a close enough approximation - by inserting a locally generated carrier and using that to demodulate the surviving sideband. Because the amplitude, frequency and phase of the original carrier cannot be known exactly, some skill on the part of the operator is required to resolve intelligible speech as opposed to garbled Donald Duck noises.

This method of transmission, called single sideband, supressed carrier (the supressed carrier part of the description is usually, ahem, supressed) is clearly unsuitable for music or entertainment, but it is possible to transmit a smidgeon of the original carrier wave in what is called single sideband, reduced carrier to which the locally generated carrier signal may be synchronised in phase.

Double sideband amplitude modulation however, survives on long, medium and short wave radio broadcasts (and in some television systems) for the sake of simplicity in the receiver. But to use it for video would be wasteful of both bandwidth and power, and would have made the design of suitable high-bandwidth, high gain receivers difficult in the early years. Equally, it would be difficult to filter out the carrier and the whole of one sideband at the transmitter without introducing distortion into the other sideband (unlike telephony audio, which contains no energy below about 300Hz, a video signal contains a dc component - 0Hz - resulting in sidebands that converge upon the carrier frequency), and it would not have been easy to design a simple ssb video demodulator for the receiver. For these reasons the first 405-line station at Alexandra Palace in London radiated dsb vision, and the early receivers had dsb detectors.

Frequency spectrum of a System I vestigial sideband transmission showing ideal receiver frequency response The frequencies marked are relative to the vision carrier. The levels and slopes of the curves are stylised for clarity.

But in time, in true British style the boffins came up with a compromise. The whole of the carrier and one sideband would be transmitted, together with a bit (a vestige) of the other. This has several advantages. The remaining sideband suffers no distortion in the transmitter and an unmodified dsb envelope detector can be used in the receiver. All that is required is an rf (or if - intermediate frequency - in the case of the new-fangled superheterodyne receivers) response that is tailored to suit the incoming vestigial sideband transmission. In this, the response near the carrier frequency is reduced such that the carrier itself is received at half strength and the response tails off as it penetrates the vestigial sideband. In this way the lower video frequencies are received in both sidebands and the upper frequencies come from the full sideband alone, albeit a little distorted by the action of the envelope detector in the presence of only one sideband.

More modern receivers with synchronous detectors do not suffer from this distortion, and the whole video spectrum is recovered from the full sideband alone. Vestigial sideband receivers of either vintage may be used to receive double sideband transmissions. This is just as well, since the cheap uhf modulators incorporated into video games, vcrs, satellite receivers etc all operate on dsb.

Vestigial sideband is also used in NTSC and PAL colour signals, though it is seldom mentioned as such. In these two standards the two colour difference signals are amplitude modulated onto two subcarriers of the same frequency with a 90° phase difference. Because of the 90° phase shift, one subcarrier is at a maximum amplitude excursion when the other is crossing zero, and this enables the two colour signals to be completely separated in the demodulator. The actual subcarriers are supressed before multiplexing with the video (luminance) signal, and about ten cycles of subcarrier, called the colour burst, are added to the horizontal blanking interval as a reference for the demodulator in the receiver.

If the full dsb signal comprising subcarrier and sidebands were to be added to the video signal it would produce a strong interference pattern all over the picture. We have seen that the amplitude of the carrier must be twice that of the highest modulating signal, and so the whole signal would have to be attentuated severely to enable it to be transmitted at all, and then more to reduce the interference level, resulting in a very poor signal-to-noise ratio for the colour information. The dsb supressed carrier signal can be inserted at a higher level, since only very saturated colours have significant sideband amplitudes - in fact the amplitude for greys is zero.

This mode is often called QAM (quadrature amplitude modulation), particularly in digital communications.

The bandwidth of the colour signals is restricted to 1.3MHz and with a subcarrier frequency of 4.43361875MHz in the PAL 625-line standard that would give an upper sideband excursion of about 5.7MHz. In Systems B/G, I and D/K the video cut-off frequencies are respectively 5.0, 5.5 and 6.0 MHz. These are nominal, built into the specifications for each standard, but often the video is cut off around 5MHz for all standards. So the colour information is effectively vsb, but since there is little energy in the removed portion of the sideband no account of this is usually taken in the PAL receiver.

In the case of NTSC however, things are much tighter. With a 3.57954545MHz subcarrier and a 4.2MHz video bandwidth in System M, only 600kHz is available for the usb of the colour signal. Some fiddling is done to ensure that one colour signal corresponds to the colours to which the human eye has least accuity - it sees them blurred in other words. This (called the Q signal) is given a 500kHz bandwidth while the other (I) is afforded the 'full' 1.3MHz. However, the reduction of the usb of the I signal to 500kHz in transmission means that either the receiver must filter the whole chrominance signal to 500kHz or perfom ssb demodulation on the 1.3MHz lower sideband of the I signal in order to avoid severe distortion caused by demodulating the vsb colour signal as a dsb one.

Television Website Bookmarks   Mike Brown/MB21/Ether.net Andrew Emmerson/Paul Stenning/405 Alive/British Vintage Wireless Society Keith Hamer David Laine/Vintage Broadcasting Richard Logue/irishtv Darren Meldrum Richard Russell Peter Vince/Barney Wol Andrew Wiseman/625 Room Bill Wright

Back To Top

Pembers' Ponderings

Compiled by Alan Pemberton
Sheffield, South Yorkshire, England
Email me