TV SIGNAL PROCESSING:COLOUR DECODING.

COLOUR DECODING

Most of the post-detector signal processing stages of a colour TV or monitor are concerned with recovering RGB video signals from the encoded CVBS signal coming from the receiver section, video- recorder or camera. The description which follows is based on the PAL system.

The CVBS signal is resolved into U and V colour difference signals by the first part of the decoder, illustrated in the block diagram of Fig. 7.2. The processes carried out here may be briefly summarised as follows:

TV SIGNAL PROCESSING-0102

1. Chroma amplifier with bandpass filtering based on 4.43 MHz to separate the chroma signal from the composite waveform and to provide initial amplification.

2. The output from the chrominance amplifier is fed to a delay line and signal separating network which separates the U and V signals by a process of adding and subtracting the direct (real- time) signals to and from those which have passed through the delay line. Here also is carried out the hue-averaging process between successive lines of chroma information.

3. The R−Y and B−Y outputs from the signal separation circuit are applied to synchronous demodulators which sample each at regular and short intervals according to the timing of their drive (reference) signals, derived from the reference oscillator. As there is a 90° phase difference between transmitted U and V signals the reference feeds to the synchronous demodulators are likewise in quadrature. In this way the upper demodulator is made conductive only for U signals, and the lower only for V signals.

4. As the burst signal is required to synchronise the reference oscillator, a burst gate is needed to separate the colour burst from the rest of the chroma signal; this can be achieved by gating the stage with a pulse from the line timebase section so that it is switched on only during the back porch period when the burst signal is present.

5. As the subcarrier is suppressed at source it is necessary to provide a local oscillator as a source of reference carrier for the synchronous demodulators. This reference oscillator must operate at the same frequency and with the same phase relationship to the chroma signals as the subcarrier itself. For accuracy and stability a crystal oscillator is used; its phase – and its frequency within narrow limits – can be ‘steered’ by means of a phase detector and reactance control circuit.

6. The burst signal switches phase between ±45° on alternate lines, giving rise to a signal at 7.8 kHz (half line frequency) appearing as a ‘ripple’ at the output of the phase detector. This, the ident signal, is taken off by the 7.8 kHz acceptor filter and used to control the colour-killer and PAL switch phase-inverter circuits.

7. The colour killer circuit, controlled by the ident signal, operates to cut off the chroma signals during reception of monochrome pictures. This is necessary to prevent spurious signals (i.e. h.f. components of the luminance signal) and noise in the chroma circuits – whose gain in these circumstances will be at maximum under a.g.c. (a.c.c.) control – from generating cross-colour and confetti effects on the picture. When the ident signal disappears the colour killer shuts down the chroma amplifier.

8. Since the PAL system depends on phase reversal of the V chroma signal on alternate lines, it is necessary to incorporate a phase inverter (180° phase shift) controlled by a bistable switch which brings it into operation on alternate lines. The bistable switch is triggered by a pulse from the line oscillator, and is locked into the correct phase by the 7.8 kHz ident signal. The PAL switch may alternatively be incorporated in the V signal feed line, where it will have the same effect of restoring correct polarity to the V (hence R−Y, and finally R) chroma signal.

Many variations in decoder design are possible, as will be seen when IC decoders are investigated shortly.

Delay line and U/V separation

The line-by-line phase inversion of the V chroma signal in PAL- encoded transmissions was described in Chapter 6. The errors introduced by differential phase distortion in the signal path remain relatively constant so that they will have opposite effects according to which of the two V axes is in use for transmission. For instance if a blue area is being transmitted it may be shifted towards cyan dur- ing one line, but towards magenta on the succeeding line. If such a signal is displayed on a picture-tube the human eye will tend to integrate and average out the opposing hues and see the correct blue hue. Large phase errors would give rise to a noticeable venetian blind effect, sometimes referred to as Hanover bars. The equal-and- opposite errors can be effectively cancelled out electronically by means of a one-line signal delay and associated matrixing circuits. Consider Fig. 7.3. Fig. 7.3(a) has a phasor A (solid line) transmitted on line n at 30°; phase distortion in the signal path causes it to be received as dotted phasor B at 40°. The same hue transmitted on line n + 1, now A´ at –30° with respect to the U axis, appears in Fig. 7.3(b) along with the phase-distorted resultant as a dotted line phasor B´ at –20°. After reversing diagam b’s vector about its U axis, the two received signals B and B´ correspond to B and B˝ (Fig. 7.3(c)), which when averaged give the correct hue A at 30°.

The essence of the decoder in a PAL receiver, then, is in the separation of the chroma signals modulated on the subcarrier, and in so processing them that the electrical average between each successive pair of lines is applied to the synchronous demodulators. It depends on the use of a delay line with which a one-line-old chroma signal can be made simultaneously available with a real time signal. The

TV SIGNAL PROCESSING-0103

delay line is made of glass; an input transducer converts the electrical subcarrier signal to an acoustic (mechanical) one, whence it is propagated relatively slowly through the body of the glass block in a zig-zag path, being reflected, snooker-ball style, whenever it encounters the glass wall. The path length is such that the transition time is exactly 63.943 μs ±3 ns. At path-end the mechanical wave is reconverted to an electric one by a piezo microphone. The 63.943 μs delay time corresponds to 283^ cycles of reference subcarrier: the odd half-cycle is very important, ensuring that the emerging signal is in opposite phase to that of the undelayed (real-time) subcarrier signal. Since the U signal is always transmitted in the same phase, addition of these direct and delayed signals (Fig. 7.4(a)) will cancel out U components altogether. V signals will be present, however, because the phase reversal introduced during encoding will effectively cancel out the phase reversal due to the half-cycle offset at the delay line’s output. The time-coincident V subcarrier cycles, then, will sit on each other’s shoulders in the adder to render a pure 2 V output. Now consider the subtractor in Fig. 7.4(a). It receives the same two signals as the adder, but here the subtraction of each V subcarrier cycle from its identical and time-coincident fellow will render zero V output. Antiphase U cycles, however, when subtracted will reinforce each other to render 2U as sole output. An earlier chapter revealed a subtractor to be a combination of inverter and adder, and they are thus drawn in Fig. 7.4(b), making clearer the separation of the U signal. Since the chroma signals of two consecutive lines contribute to all U and V outputs the required averaging outlined in Fig. 7.3 is achieved simultaneously with the separation process.

TV SIGNAL PROCESSING-0104

Synchronous demodulation

Every NTSC or PAL decoder contains synchronous demodulators which are basically on-off switches, closed briefly once per subcarrier cycle by the reference waveform. The timing of the ‘on’ period is governed by the phasing of the local reference feed, which is locked in the required phase by the burst signal via a phase-locked-loop (PLL). Because the U and V subcarrier signals are in quadrature the peak of one coincides in time with the passage of the other through the zero line. Referring to Fig. 7.5, it can be seen that closing the U switch at times t1 and t2 will sample the U signal without crosstalk from the V signal, and that closing the V switch at time t1 and t3 will likewise sample only V level. The U and V signals, of course, represent colour-difference signals which can have either positive or negative values – on a yellow subject, for instance, R−Y and G−Y will take the form of positive voltages to turn on the red and green guns of the picture-tube, while B−Y will take up a negative voltage to turn off the tube’s blue gun. Study of Fig. 7.5 shows that the synchronous demodulator is capable of producing positive and negative outputs, and in practice each demodulator’s output varies rapidly in terms of amplitude and polarity as the chroma subcarrier signal (there is only one in spite of the diagrams of Figs 7.4 and 7.5) changes its phase and amplitude to describe the hue and saturation of the picture elements in turn.

TV SIGNAL PROCESSING-0105

Crystal oscillator

A resonant crystal consists basically of a tiny slice of quartz mounted between metallic plates, and enclosed in a sealed envelope. Like a tuning fork the quartz slice has a mechanical resonant frequency; unlike the fork, it is barely affected by temperature variations, and behaves like an extremely high-Q tuned electrical circuit. While its stability is very good, it cannot by itself provide a reference signal of correct frequency and phase. Fortunately the resonant frequency of a suitable crystal can be slightly ‘pulled’ by capacitive loading; the use of a vari- cap diode for this purpose permits voltage control of oscillator frequency. This combination forms a voltage-controlled crystal oscilla- tor (VXO) and is used in combination with a phase detector to make a much-used building block in electronic circuits – the phase-locked- loop. Its principle was described in Chapter 3, and here the PLL is used to lock the local crystal to the mean phase of the transmitted burst signal. To prevent the local reference crystal trying to follow the swinging phase alternations of the PAL burst signal a suitably long time constant is present in the error voltage path.

G−Y and RGB matrixing

Before a G−Y signal can be made from the red and blue colourdifference signals the transmitted U and V chroma components must be de-weighted. It was explained in the last chapter that a reduction

in amplitudes of R−Y and B−Y signals is made to prevent them from overdriving the transmission system. After passing through an amplifier with a gain of 1.14 the V signal is ‘normalised’ to R−Y. Similarly, B−Y appears at the output of a ×2.03 amplifier fed from the U signal. Recovery of the G-Y signal depends on the basic equation given earlier: Y = 0.59G + 0.3R + 0.11B. In fact, G−Y can be directly derived from R−Y and B−Y. Adding 0.508 of (–R−Y) to 0.186 of (–B−Y) renders G−Y. Correct proportions of inverted R−Y and B−Y signals meet in an adder and combine to render the G−Y signal. Finally, the three colour-difference waveforms are each added to the luminance signal to make the primary-colour signals R, G and B for application (in inverted form) to the picture-tube cathodes.

Because the bandwidth of the luminance channel is kept wide to ensure that a detailed black-and-white picture is displayed as a base on which the coarser colour information is superimposed, the transit time of the Y signal is much shorter than that of the chromi- nance signals, constrained in a channel about 500 kHz wide. When the Y and colour-difference signals come together in the RGB matrix this would result in them being out of step to cause misreg- istration on the screen. To prevent this a short delay line (t about 500 ns) is provided in the low-level luminance path. The glass delay line described previously is not suitable for a wideband signal; the type used here has series inductance and parallel capacitance distributed along it, and takes the form of a low-inductance coil wound over a grounded foil. Alternatively, a bucket-brigade device (of the type described in connection with CCD image sensors in the previous chapter) is used, with its advantages of small size and IC construction.

Subcarrier trap

On highly saturated colours a large amplitude subcarrier signal is present and appears on the screen as a fine dot-pattern. Although the pattern itself is barely distinguishable at normal viewing distances the non-linearity (gamma) of the tube will have the effect of partially ‘rectifying’ this subcarrier signal, artificially brightening up highly coloured parts of the picture. A notch filter, sharply tuned to 4.43 MHz, forms a trap in the luminance signal path; in some decoder designs provision is made to switch off this trap during reception of monochrome transmissions, thus realising the full definition capability of the shadowmask tube. The trap-switching is carried out by the colour-killer line.

Leave a comment

Your email address will not be published. Required fields are marked *