Wireless World, Seplember 1971 443 Field-sequential Colour Receiver

I-Introduction and basic principles

by T. J. Dennis, B.A.

All systems of colour TV in general use higher than the flicker-fusion rate of the normal broadcast colour transmissions. today have as their display a system eye, a highly variable quantity found on A standard PAL decoder4 provides whereby the three primary coloured average in the region of 30Hz. the three (narrow band) colour difference pictures are spatially superimposed, For this reason it is normally considered signals. These are then switched in turn whether by projection of the , and necessary to increase the basic field rate to the grid of a monochrome c.r.t., the images using the Schmitt system, by from that of, say, a 50 field monochrome change taking place during the field blank­ the use of three c.r.ts and half-silvered standard to 150 fields per second in order ing period (see Fig. 1). The luminance mirrors or with the three pictures on one to maintain the original luminance rucker (wideband) signal is fed to the c.r.t. cathode c.r.t. whose screen consists of triads of rate. This demands a trebling of the signal as usual, after its passage through a 60005 independently controllable dots, bandwidth, other factors being constant. delay tine. This is practically the only as in the R.C.A. Sbadowmask' tube. Noting the discouraging comments of major modification needed to the mono­ All three systems are capable of others on the subject of f.s. systems retain­ chrome receiver which is the source of all J excellent results, but are difficult and ing the existing monochrome field rates , the signals used. The net result is that the elCpensive to set up. For example, in the it was decided to attempt to bUild and set can be made to display, field sequen­ projection system complex distortions have operate such a unit, to work from the tially, the black-and-white equivalents . of to be introduced into the scanning wave­ forms to correct for the fact that the ' Goldmark, P. C., et aI., ' ', Part I, "Sec T. D. Towers on principles of colour TV, lan.­ projectors cannot be co-sited. Much the Proc. I.R.E., Vol. 30, pp. 162-182, April 1942. Dec., 1967, Wireless World. same problem is encountered with Shadow­ mask tubes, hence the joys of convergence adjustments. Any system using separate electron sources is prone to grey-scale l.um. ------;~~ C Q'i~·~d., tracking errors. The Shadowmask is able to reproduce G-Y Line a range of colours because the spatial flyback puiS., colour resolution of the eye is poor: close to a screen the dots can be easily perceived, but the overall impression is still one of the 8-Y B uffer Chromo c.r.t . additive colour resultant. Temporal colour amplifl.,r output grid resolution is equally weak, as may be seen by rotating a disc carrying segments of, 4 5 say red and blue, when the colours rapidly merge to magenta as the speed of rotation is increased. This is the basis of the R-Y field-sequential process, whereby the three coloured images are presented to the eye in turn. It is the oldest form of colour display a version having been demon­ 2G strated by J. L. Baird in 1928.2 In the author's opinion it is capable within its limitations, of giving results of the highest quality. Coil Perhaps the major of these limitations 12 is caused by the eye itself: perception of luminance, or brightness, changes in time, as well as in space, are particularly good. 50Hz Colour input While a rotating disc of red and blue will wh.,el appear magenta, it also carries a marked brightness rucker due to the luminance +12V difference between red and blue. Flicker only disappears when its frequency is

6 Coarse­ 'Herold, E. W., 'Methods suitable for TR color vel. 1) kinescopes,' R.C.A. Review, Vol. l2, Sept. 1951, P 445 et seq, " TV in natural colours demonstrated', Radio Fig. 1. Block dlagram offield-sequential colour receiver eqUipment. The input is provided News. Vol . 10, p. 320, October 1928 by a normal monochrome receiver. 444 Wirele s World, September 1971

the red, green and blue images of a colour on the screen (not per field). when display­ transmission. ing any but saturated primary colours, will A disc carrying sections of primary red, differ in shade. However since the colour green and blue filters rotates at 16jr.p.s. detail resolving power of the eyc is poor, in front of the C.r.t. Its rotation is phase the eJTect could only be seen within about locked to the field sync pulses to ensure 12in. of the 14in. C.Lt. used. Bearing in that when the red picture is being scanned, mind that these initial tests were on 405 the red filter is in place, and so on. lines, with a 625-line colour picture at lt may be noted that two-thirds of the normal viewing distance, the effect is available colour information is wasted in unnoticeable. _ this system, but it should also be recalled After the encouraging results obtained that considerably more than two-thirds of with the colour bars. a PAL decOder was the energy imparted to the electron beams built, with slight modifications notably in in the Shadow mask tube is dissipated as elimination of dependence on the line out­ heat in the Shadow mask ! put stage of the receiver: an additional Phase lock of the colour wheel is re­ sync separator was added, the line pulses quired, to ensure that the correct filter is obtained being used to trigger a mono­ in place at the correct time. This is achieved stable and produce an accurately timed by a simple feedback system using a signal burst gating pulse. The burst gate itself derived from a coil wound on aU-shaped was in the fonn of a four-diode bridge, all transformer limb and mounted in front of of which will be discussed more fully next a bar magnet fixed to the centre of the wheel. month. The coil output waveform is square in At first the decoder was operated without form, with slow sinusoidally changing a delay line; i.e., in the PALoS mode. edges. A foUr-diode bridge is used to gate Oscilloscope examination of R - Y for the through an 8ms portion of this wavefonn, colour bars with careful adjustment of L6 16G. AI. 'I. in. AI. which has a manually controllable d.c. of the May 1'969 W. W. article enabled potential superimposed on it for coarse results to be obtained which did not differ speed adjustment. The bridge output is appreciably line by line. Stability over integrated, amplified, and with suitable d.c. long periods, however, was not good due level adjustments, used to drive the motor to mechanical vibration and thermal armature via a 2N3055 emitter follower. changes. Hanover bars were then obtained. Assuming phase lock, the gating pulse Addition of a PAL delay line effected a is placed symmetrically about the mid­ complete cure. Adjustment of the R - Y, B - Y and point of the positive going edge of the Bar feedback waveform. H the motor speeds magn .. t G-Y drives to the sequential switches up for any reason, the waveform reaches a enabled colour pictures to be obtained higher level than it would normally when whose fidelity was indistinguishable from sampled, and the integrator output moves Shadowmask results, with the advantages of in a positive direction. Because of the full luminance bandwidth (a notch mter has inverting amplifier, the armature voltage is been found unnecessary; some commer­ reduced, and the motor slows down. By cial receivers do not include them), similar reasoning it can be shown that a and total elimination of the necessity for reduction in motor speed will also be com­ complex convergence and grey-scale pensated. Not surprisingly, the sy tern tracking adjustments. With the latter even oscillates about its stable position when any 5in. if the filters do not give an exact white, velocity transient is applied; settling time there can be no failure, since the same gun from switch-on is about 20 seconds in the G .. , ati n .. filter is used for all three pictures. Problems prototype, but this is immaterial as it takes will arise, however, if any attempt is made the line timebase considerably longer to to provide switched compensation for warm up on the displaying set. Programme filters of wildy incorrect characteristic. switchings, when field sync may be'inter­ As mentioned above, field-sequential rupted, tend to upset phase lock, but this systems working at low field rates suffer effect has not been found troubleseome. V.in. from luminance flicker effects. Another problem is colour fringing, obtained when there are differences between adjacent Results Fig. 2. Front view (a) and section (b) of a simple cowur wheel. fields' i.e., when the scene contains Before embarking on the construction of movement. a PAL decoder, a generator was built to Fortunately, both have proved a far less produce the 4f, 2f and f, where f = line serious drawback than was expected. frequency squarewaves needed for the least well presented as on all colour sets. Perception of flicker depends on many blue, red and green, respectively, signals The reddish tint obtained when observing factors including background light level, of the standard colour bars, viz. white, a white object through the wheel (due degree of dark adaption and size of the yellow, cyan, green, magenta, red, blue and to the red filter having excessive trans­ field under consideration. Thus, viewing black. These were applied to the linear mission) was neatly compensated by the a f.s. picture under well lil conditions gates. The resulting non-eomposite out­ blueish tint of the c.r.t. phospbor. results in the fiicker being highly objection.­ put was then passed through an existing Passers-by who ventured unsuspecting able: lhe colours are desaturated, and may camera channel, and emerged with a full into the lab. during this stage of develop­ not be seen at all. This seems to be true set of 405 line-standard sync pulses, for ment were invited to peer through the disc, whatever the brilliance of the displayed ease of application to a monitor. and report the colours seen. Most were image, which has in any case to be high to The resulting wildly flashing vertical correct without prompting, but two insisted overcome the effects of reflected light from stripes, when viewed through a locked they saw bLue and red separately on the c.r.t. screen and colour wheel. colour wheel, became the familiar bars. magenta bar. This onJy tended to happen The improvement when pictures are Colour fidelity, even with the rather crude at high brightness levels, and is an effect viewed in either total darkness, or very ex-stage lighting filters in use, was in not observed by the author. low ambient lighting is considerable, general excellent, the yellow being the Owing to interlacing, successive lines particularly once dark adaption has taken Wireless World, September 1971 445 place. Flicker due to the luminance differ­ Direction 0 r rotatIon Fig. 3 shows the basic form of the spiral' ence between the red, green and blue images ------... holed wheel. while the calculations for the in a black-and-white transmIssIon is diameter of it for given raster size are negligible, while there is no sensation of illUSl rated in Fig. 4. colour at all. Let the dimensions of the raster be 4x by [n general, flicker in coloured pictures 3x, assuming a standard 4:3 aspect ratio. increases with increasing area of colour, Then in triangle OAB, its saturation and luminance level, and is OBl=il greater for the primary colours, particularly = AB2+A02 green, than the complementaries. The latter = 4r+ (3x+ y)2 is true, since the mark-space ratio with the = 13x2+ 6xy+.yl, whence, by taking saturated primary colours is 1:2 (i.e. one the square root, r can be determined. field out of three is displayed) while for The dimension y is determined by the saturated cyan, magenta and yellow, this physical size of the driving motor, and ratio is 2: I. other factors such as mounting arrange­ MOSl programme material does not, ments. but a useful rule for a minimum however, carry large areas of saturated value is to make y= one-third picture width. colour, and the viewer may be unaware The radius required for this type of of nicker, depending on the content of the disc is clearly less than that for the simple programme and its degree of ' viewer wheel first described: Fig. 3. Spiral colour wheel. The 2 involvement'. spiral cut-Ollts are represented by the For a simple wheel, radius = 18.25x2 A warning is due here: it is probably position of scanning lines seen through + 8X;' +y " by similar reasoning to that unwise for anyone susceptible to flicker, as the disc at 1/12thjield intervals. This above, compared with ,-l = 13x2 + 6xy +y­ in some cases of epilepsy, to view colour figure is drawn for a raster of for the spiral disc. television in this way, as it contains, as dimensions 6 X 4.5 units, the central well as major components at 16jHz and area having a diameter of 4 ullits. 50Hz, smaller components at 25 and stHz due to interlacing. The latter par­ ticularly is close to' the so-called danger frequency of 7Hz. However, the author relation to the pickup coil is adjusted so who .does not suffer from epilepsy, has that the required section of filter moves used himsdf as guinea-pig in viewing down the c.r.t. with the Reid scan carrying trials as long as three hours, with no ill that colour giving a maximum segment effects, apart from a crick in the neck from of the wheel through which the correct the difficult viewing position necessary colours can be seen. This implies that the with the prototype: the colour wheel is colour picture can only be viewed with one lOin. in diameter, and close to the eye eye through the side of the disc; however, if while the raster is on a 17in. c.r.t. 4ft away. the viewer moves back aboUl two feet, The second problem of colour splitting the right-hand side of the picture can be is, of course, only apparent on images seen with the left eye and vice versa, with carrying movement; it has. however, been only small (top right, bottom left) areas found that any movement has to be quite cutofl'. An alternative fonn of disc uses spiral fast before splittin~ becomes visible, the gesticulations of an orchestral conductor areas of colour, which follow the field scan being particularly susceptible. In most down the screen, and enables the colour case , though, the subject of attention in a picture to be seen th rough the top of the scene is kept stationary on the screen, wheel. The spiral wheel can be made slightly while the background moves. An example smaller than its simple counterpart, thus a specimen for operation directly in front of a is a horse-race, where the rails can be seen by a conscious etTort as red, green 10 or II inch c.r.Us feasible. and blue bars. Spirally cut colour dmm in front of the cathode-ray tube.

Possible forms of colour wheel Perspex is an ideal material from which The prototype colour wheel was a simple to fabricate a spiral wheeL, as the diameter affair, and is shown in Fig. 2. can then be made the minimum possible. It Two aluminium discs were cut out using is very easy to work, the routing machine a woodworker'S routing machine. One was again being ideal for cutting out the two *th F n. thick, the other 16 gauge. The three inch thick discs required. The discs are cutouts were then made with the same fitted to an identical mounting boss to that tool, which proved remarkably efficient a used with the simple wheel. but care should bolt being placed through the centre of be taken with clamping bolts, as the plastic the disc and router plate so that tbe cutting tends to shatter under pressure. Aluminium edge of the router was at the required discs of radius dimension y should be radial distance. The diso was then rotated placed on each side of the Perspex at the slowly, leaving the radial arms of the wheel. centre to spread this load. Periphery clamp­ Straight sections were cut with a hack­ ing screws should be countersunk 6BA saw. Pieces of red, green and blue gelatine types, and no Longer than necessary, to Rlters (as used for stage lighting) were minimize windage. Again. they should not sandwiched between the discs, which were Fig. 4. Disc radius for given raster size. be overtightened. clamped together by the screws through the Spiral colour wheel mounted ill fr01ll In order to obtain a 9,6in X 7.lin (l2in machined mounting plate, and 6BA screws of the c.r. tube. Rotation is anric/ockwise. diagonal) colour picture, a 23 inch diameter into holes tapped in the 'in. disc The oUlomaric phase-control components specimen of this type has been manufac" periphery. are mounted at the rear of the equipment tured· with uccessful results. Careful The angular position of the magnet in on the colour wheel shaft. balancing of a wheel of this size is 446 Wireless World, September 1971 necessary, and this was carried out by placing it, with a 6in length of shaft through the centre, between two horizontal edges, Voltage Reference Source and adding pieces of lead to the screws through the protective aluminium discs, until the wheel would remain stationary in Constant-current drive with 0.08% stability any angular position. The following instructions, in conjunc­ tion with Fig. 3, can be used to construct the spiral holes. The figure can be con· by H. A. Cole*, M.I.E.R.E. veniently drawn on the protective paper covering the Perspex sheets at purchase. 1. Calculate the desired radius of the disc from the selected values of x and y. (x can be determined from the relation 5x= diagonal of raster used, since the diagonal and two of the sides ofa4:3 raster make the '34-5' triangle of elementary geometry.) 2. Divide the height of the pictUre into­ say-twelve sections of length a (i.e., Specially constructed zener diodes having a= 3x/12). very low temperature coefficients (less than 3. Divide a 1200 segment of the disc into + O.OO2%/deg C) are now readily available D3 1 the same number of segments, here twelve ______VD3 of ten degrees eacb, drawing radial lines. it moderate cost, and are intended for ,.. ~1-N-8-2-5~r_o7·5mA applications in which a highly stable voltage reference is required. However, unless the operating current of such diodes is maintained within closely defined limits, D, the advantage of a low temperature c0- OAZ242 efficient will be lost due to voltage variations occurring across the internal impedance of the diode. There are many ways in which a constant Tr2Q operating current can be provided for a "107003 reference diode but one of th.c simplest and most effective method is by use of the ring-of-two' circuit introduced by Tr Williams in 1966 (references 1). Unfortu­ 1b nately, although this circuit performs "107002 extremely weU over a wide range of supply voltage variations, its inherent temperature D2 dependence (about - 4mV ldeg C) makes it OAZ242 unsuitable for use in circuits subjected to wide variations in temperature. The princi­ pal cause of its high temperature L------~~-30V dependence is variations in Vbe of the Synchronizing magnet and pick-up coil. two transistors (typically - 2m VIdeg C each). Fig. 1. To compensate for temperature At first sight it might appear feasible sensitivity in the transistor V be of the 4. Draw a line of length 4xperpendicular to compensate for dVbJdT by selecting ring-o/-two voltage reference, dual to and bisected by, the radius pointing zener diodes used in the ring with tra1lsistors with matching V be are used. to~ards the top of the paper, at a distance temperature coefficients identical to those of One base-emitter junction is used as the 12a+ y, (i.e., height of picture + y), from the transistor Vbe- UnfortunaJ:ely, although compensation diode for Vbe variations of the centre of the circle. such an arrangement is not impossible, the the other transistor. From Fig. 2, circuit 5. Repeat step (4) with each radial line, difficulty of obtaining a zener diode having is required to give 7.5 ± O.2rn..tf through moving in an anticlockwise directio~ and the desired voltage and temperature c0- D]. reducing the distance of the perpendicular efficient (of the desired sign) is consid~able . from the centre by length a each time, A better solution is to use zener diodes until the 13th radius is reached, when a line which have a negligible temperature 7.5mA by the remainder of the circuit distance y from the centre should be drawn. coefficient and then connect an ordinary 6. Repeat steps (3) to (5) for the remain­ connected to operate as a ring-of-two. each forward-biased diode in series with each. ~ ing 1200 segments of the circle, starting half providing 3.75mA (Fig. 1). The ref­ arrangement like this lends i~elf rea~y where the innermost perpendicular of the erence voltage for each half of the ring is to the use of a dual transistor WIth matching formed from the series connection of a previous spiral was drawn. Vb as the base-emitter junction of one This process builds up an envelope of tile ~ . zener diode having a very low temperature transistor can be used as the compensation coefficient (typically +O.SmVideg C at spiral holes needed, which can be com~leted diode for variations in V be of the other 3.75mA). and the base-emitter j un cti~n of freehand . Using a great.er number of lUCre­ transistor. Unfortunately, because of rnents will, of course, increase the accuracy, a transistor operated as a forward-blased unequal currents in the two junctions, diode. The overall temperature coefficient but tend to clutter the diagram somewhat. complete compensation cannot be expected. Slices of the coloured filters should be cut of each series connection is about - J .5mVI A circuit based on this arrangement. but deg C and provides reasonable temperature to shape, and sandwiched between the ~cs using transistors with unmatched base­ after removal of all paper but that carrylDg compensation for variatio~s in V be ?f the emitter voltages, is now described. transistor which it supplies. In Fig. 1, the design. Small pieces of adhesive tape can A high-stability reference diode-D]­ therefore, the zener voltage of D I ' plus the be used to secure the filters in position dur­ is supplied with a constant current of ing final assembly, after which all areas of Vbe drop of Trzb forms a temperature­ the wheel needed to be opaque should be compensated reference voLtage for the coated with blackboard paint. • A.E.R.E., Harwell transistor Tru' In a similar way Tria (To be concluded)