40 PHILIPS TECHNICAL REVIEW VOLUME 27

The effect of errors colour yellow, for example, will be alternately too What, then, are the technical reasons that have leadgreen in one line and too orange in the next. A large a number of experts to prefer the PAL or SECAMarea of colour, however, seen from far enough away, system to the longer established, further developedwill nevertheless give the right hue, since the ratio and, as practice has shown, quite satisfactory NTSCbetween R'-Y' and B'-Y' is on average exactly right. system? The main reason is the concern about theThe only ill -effect will be a degree of saturation that phase modulation of the NTSC sub -carrier. As we haveis slightly too low (since cos a < 1), and this in general already said, the phase of the sub -carrier in the NTSCis not very troublesome. system determines the hue. What can go wrong with The above sets out the basic principle of PAL, but this? A phase shift of the complete NTSC sub -carrierunfortunately, in practice things may be a little differ- signal cannot in itself give trouble, since the hue isent. Because of the non -linearity of the picture tube determined by the phase difference from the referencecharacteristic (y 0 1),a 'phase error gives rise to signal (the burst) which is also present in the signal,brightness as well as colour errors and the brightness and not by the absolute phase. Only if the phase shiftserrors are also alternate on successive lines. This for burst and signal are different willgives a fairly coarse pattern of stripes which moreover colour errors occur. Such in the signal canappears to move in the vertical direction, because of arise in certain unperfected equipment because thethe interlacing. This effect has the result that the sensi- chrominance signal is superimposed on the luminancetivity of the PAL receiver to differential phase errors signal, while the reference signal always is given atis as great as with NTSC, although the effects on the (see fig. 5). If a phase shift dependent onpicture are quite different. In NTSC they appear as the level now occurs, there will be colour errors in thecolour errors, but in PAL as a moving pattern of NTSC image. stripes ("Venetian blind effect" or ""). Although it has been found quite easy to keep these Nevertheless it is possible to improve the perform- "differential phase errors" to a minimum with modernance of PAL receivers in this respect. Instead of equipment, a more attractive idea appeared to be to make the system insensitive to such errors. For both R=Y" PAL and SECAM, the basic concept was the avoid- ance of difficulties due to phase shifts dependent on the level. However, "what you gain on the swings you lose on the roundabouts", and so this can only be done at the expense of something else. In the SECAM system, the desired result is achieved by the, use of : in fact in areas where colour is uniform the frequency remains constant and the Fig. 15. Vector diagram of the phase has no effect. Differential phase errors can give chrominancesignalof PAL rise to colour errors only when the brightness varies, for two successive lines, where there is a differential phase error but this effect is only slight and rarely visible in practice.a. The vectors shown in broken The price that has to be paid for this advantage haslines occur instead of the sol- idly -drawn vectors C. Because already been partly explained above: the possibilityq' changes itssign each time, of annoying errors at vertical colour transitions, andwhile a does not, there is a cer- tain amount of phase error com- poorer compatibility. pensation in PAL. In the PAL system, differential phase errors are in the first instance less of a hazard because the colourleaving the averaging out of errors to the eye, a decoder variations are averaged out. If the chrominance signalcan be used which does this for the viewer by directly contains a differential phase error a, then instead ofaveraging the electrical signals. This can be done with the desired signal, e.g. C sin q,, the circuit shown inthe aid of a delay line, of the same type as that used fig. 12 gives the signals C sin(g)-F-a) and C sin(q)-a)in the SECAM receiver, giving a delay of one line for alternate lines in the (R'-Y') direction, see fig 15.interval (64 t.ts) (see fig. 16). The difference between On average, thisis-1-C [(sin(g+ a) + sin(q)-a)] =this scheme and SECAM is, of course, that the delay C sin q, co's a. Likewise, the result will be on averageline is essential to the operation of SECAM, whereas C cos q cos a for the (B'-Y') direction. The colour isit is introduced here purely for correction. The simple therefore incorrect for every line, because the colourPAL decoder is now known as PALs, that with a compon'ents have the phase .9) + a instead of 9), butdelay line being referred to as PALd. the alternate colour errors are in opposite directions. The If the PALd decoder is used, the receiver is not very