Most engineers in the TV receiver industry will have been familiar with the Philips G8 colour chassis which was been produced in very large numbers since its introduction in 1970. The G8 has of course been up-dated in a variety of ways since then, but the basic concept, and the policy of making the up-dated plug-in units interchangeable with their predecessors, have been carefully preserved.
The time had come however for a completely new chassis, and Philips have duly come up with the G11. The G8, and its 110° derivative the G9, will continue in production for some time but the G11, with its 20AX 110° in-line gun tube, is clearly intended eventually to take its place and to have a long life.
An important new chassis such as the G11 is not an everyday event, and it’s of some interest therefore to see in detail how Philips tackled the difficult task of evolving a new chassis which is intended to be the basis of their colour television range well into the 1980s.
The Commercial Brief
The operation started with a commercial brief which consisted of a detailed specification of the performance expected of the new receiver and other technical details, together with a long list of preferences and ideas. Some of the more important features set out were: the chassis to be used in a range of receivers with different tube sizes; must be adaptable for use with a variety of styling presentations; must provide good picture and sound quality; flexibility required to allow for different forms of customer controls; a very high degree of reliability; lower production cost; even greater ease of servicing than in the G8; a long life; a high standard of production quality; and full compliance with the safety requirements of BS415, the European specification IEC65, and the radiation limits of EEC directives.
In addition it was considered important to keep to the policy of using interchangeable plug-in assemblies in order to allow future design evolution rather than revolution. This enables continuity of servicing and production techniques to be assured for a considerable time ahead.
This list of commercial requirements is clearly a formidable one but has now been translated into a product which is being mass-produced on a highly streamlined assembly line. Only time and the customer however will tell whether Philips have got it right.
The Engineering Concept
Such a complex product as the G11 was not dreamt up overnight by an inspired engineer whilst having his bath. Numerous discussions between the commercial, design, styling and production departments were required, the accumulated facts and ideas obtained from many design studies spread over quite a long period being used as the basis for the final decisions. Experimental receivers were built and discarded, but the experience gained in these exercises was stored for future use. Finally an overall concept emerged from what began as a jumble of competing, and sometimes conflicting, ideas.
To start with it was decided to use the 110° 20AX range of display tubes. These give good picture quality combined with self-convergence while the yoke system requires only minor adjustments to take into account manufacturing tolerances. Furthermore the overall depth of these tubes is small: if this could be exploited by the engineers it would reduce the limitation imposed on the styling department. A slimmer and more attractive cabinet in a wider variety of presentations became possible.
The requirement for a slim cabinet posed a considerable challenge to the engineers however. How could the electronic components be housed in this awkwardly shaped space in a manner which still met the need for the greatest possible ease of access for servicing?
At this point another factor had to be considered. It was important not only that virtually every component in the G11 should be part of an assembly which could be easily removed by unplugging, but also that it should be mounted on a printed board. This would ensure that nearly all soldered joints could be made using automatic dip or flow soldering techniques, with a consequent improvement in reliability and cost. Saving in screw fixings and other forms of mountings would also save cost, and contribute towards a more integrated and tidy structural form.
The next decision to be made concerned the possibility of using complete modular construction. At a quick glance the idea of using small plug-in modules is highly attractive think of computers and certain types of professional equipment. Then think of a plug-in module containing a line output transformer and the high voltage and high current circuits. How do you split it up into genuine small modules?
To test the concept, an experimental working modular receiver was built and design studies were carried out on variations of this approach. It soon became clear that a fully modular system involved an enormous number of interconnections and a complex and bulky, structure. The implications in terms of poor reliability and high cost were unacceptable. Furthermore it was difficult to fit the numerous modules into the space available.
The next possibility to consider was a part modular approach. This involves a mixture of modules and individual components mounted on a mother board. It has been done before by several European set makers. The drawback is that although the modules are easy to service, the mother board is not. It is not a convenient unit for unplugging and exchange or repair on the bench. So this idea, too, was not acceptable.
After a long programme of development it became clear that the best approach consisted of using modules considerably larger than conventional ones but smaller than the assemblies used in the G8. Each module would be functionally complete, and independent as far as possible for ease of assembly, testing, and servicing.
By this time a circuit diagram had been drawn up for the whole receiver, incorporating all the electronic decisions generated as a result of earlier development work. From this it was possible to establish the total printed board area needed, also the best ways in which the circuitry could be divided up, and hence the possible permutations of different numbers of boards and their sizes. In point of fact it very quickly became obvious that the circuitry divided itself naturally into certain groups of functions. These needed only inputs and outputs of a basic nature, and also resulted in boards of a convenient size — small enough for ease of assembly, servicing, soldering, handling, testing and alignment, but large enough to make economic sense and to avoid a proliferation of interconnections.
Further study showed that although the necessary printed board area could be housed in the cabinet in a variety of ways, in most cases the accessibility was not good. There was the added difficulty that some arrangements demanded board sizes that did not suit the natural circuit split established earlier. It finally became clear that the best configuration consisted of boards placed in a vertical plane behind the tube. Calculations of board area showed that this was indeed practicable, but that the layout of the components and the design of the printed copper conductor patterns would have to be carried out by skilled engineers in order to make best use of the space available. The choice of layout seemed good, but it was not an easy one to put into practice.
Clearly the plane surface behind the tube could be divided into a wide variety of board sizes, and these could be mounted in a number of different ways. After several brainstorming sessions by a group of engineers, agreement was finally reached on a basic principle: the boards would be housed in two doors mounted one each side of the tube neck. This enabled each of the two large board areas to be divided up in a manner to suit the circuit function split.
It was at first decided to hinge the doors by their outer edges to the sides of the cabinet. It was then realised that although ease of servicing was good, the proposal had a serious defect. This concerned the RGB leads between the decoder and the tube cathodes. Full drive on a contrasty picture involves a peak-to-peak signal swing approaching 100V at up to 5MHz. If the doors were to be hinged at their outer edges the RGB leads would have to be well over a foot long to enable the door containing the PAL decoder to be fully opened. When closed, it would be very difficult to prevent this lead from passing close to the signal circuits and inducing signal voltages in several sensitive areas. Furthermore radiation of high order harmonics in the u.h.f. band would be picked up by the tuner, causing direct picture interference. In addition to this the interconnecting leads passing between the pivot points of the doors would also be long, and would have to trail across the floor of the cabinet.
One answer to this problem would have been to place the RGB output stages on the tube base mounting board, in order to keep the high voltage cathode connections short. Unfortunately this would have made the board too large, while its weight would have exceeded the tube maker’s recommendations.
All these difficulties were overcome by hinging the two doors adjacent to the tube neck. The length of the leads was thus kept to a minimum, and subsequent experience has shown this decision to have been a particularly good one.
The locations of the principal circuit functions were fairly easy to decide. The i.f. strip and the field timebase had to be kept well away from the line timebase to avoid problems caused by the pick-up of line flyback pulses. In addition the tuner had to be kept cool to reduce oscillator drift, and the PAL decoder needed to be close to its signal source on the i.f. strip. These considerations, coupled with the natural routing of various leads, gave rise to the layout shown. This remained unchanged throughout development and was fully justified by experience.
The basic form of the G11 had now been established and this enabled detailed mechanical design work to go ahead in preparation for the layout of the printed boards. The final design of chassis structure is shown in the accompanying illustrations. Note that full use has been made of the available space by mounting a mains input board with fuses underneath the tube neck, and the preset convergence tolerance correction board above it. This arrangement is intended to allow easy access in the factory for final convergence adjustment. It also allows easy checking of the mains fuse during servicing, whilst keeping the live mains components recessed and out of harm’s way. It has a metal cover for added safety.
The Interconnect System
The interconnection system presented engineering problems of a more detailed nature. An important feature of the commercial brief was that it should be easy to remove each module by unplugging. Now this sounds quite a simple matter, but to do it properly so that the hundred odd connections do not adversely affect the overall reliability is in fact surprisingly difficult. Remember that each plugged lead has three interfaces: lead to contact, contact to contact, and contact to lead.
Following normal procedure, a detailed specification was drawn up describing the current carrying capability required, the resistance to voltage stress, push-on and pull-off forces, and many other parameters. This was followed by a survey of the interconnection systems available on the market, both in this country and abroad. A number of these were tried but none combined all the features needed.
As a simple example, one popular system made by three different manufacturers in three different countries had a serious fundamental design defect. It consisted of a pin and a socket with two leaves. These leaves were so short that they provided inadequate spring action. Any reduction in the quality of the material, or any sideways stress on the contact, would allow deformation of the leaves and hence an unreliable electrical contact.
After considerable investigation and many experiments it was considered necessary to commission a new inter-connection system, based on well proven technology, from a firm which specialises in this type of product. The key feature of this system lies in the form of the socket. The pin is a conventional one of generous size, and it slides between a fixed abutment and a long folded leaf spring which is formed from the same piece of metal. A long wiping action with a large area of contact is achieved, giving high current carrying capacity with low contact resistance. The long folded spring has a high degree of resilience, and this avoids stressing the metal beyond the yield point — with consequent permanent deformation. The same contact has been used throughout the receiver, and a useful degree of standardisation has been achieved by keeping the variation of plastic housings to a minimum.
A by-product of this philosophy is that the sixteen-way plugs (carrying the pins) are mounted on the boards and serve to reinforce the edges as shown in the illustrations.
Note that each socket (on the ends of the leads!) has a plastic indexing pin for unambiguous assembly. It also has a handle which makes unplugging easy and avoids stressing the leads.
The same engineering process of component specification and design was needed for two other items, the focus and tube first anode potentiometer units. As a result both have an exceptionally stable electrical performance in order to. prevent any changes of spot quality or grey scale over long periods of time.
General Receiver Development
Once the form of the G11 had been decided, the next step was to prepare outline drawings of the complete structure. From these it was possible to establish the exact size of each printed board and the restrictions imposed by fixing clips, interconnecting plugs and clearances above the board surface. In some cases contour maps had to be drawn showing how the overhead clearance varied at different parts of a board. This information provided a mechanical frame of reference which the circuit engineers used when designing the component layout and the copper track patterns.
The process of designing a printed board involves a multitude of electrical and mechanical problems, every one of which must be solved if the board is to be completed. The final result nearly always has a very definite character of its own, and it needs little more than a quick glance to establish its place in the spectrum between a work of art and a dog’s breakfast.
As an example of the “occupational hazards”, it will usually be found that the tallest component can be placed in only one particular spot for mechanical reasons, but must be located at the other end of the board for electrical ones. This, in its most elegant form, is known as the law of cosmic cussedness. It applies equally to input and output connections which obstinately refuse to group themselves at the required edge.
Other problems involve hot components, predetermined earth path requirements, clearances round high voltage conductors, avoiding the use of jumper wires, making the board easy to assemble for mass production, and in general persuading the circuitry to fit an unnatural rectangular board shape without wasting space. All these, and countless others, are routine problems which had to be solved in the G11 before complete prototype receivers could be made for testing.
It is a common misconception that once the individual circuits have been designed all that remains to be done is to put them all together and the job is finished. Surely this is especially true in the case of integrated circuits, where much of the design work has already been carried out? In practice this has never been the case, and the advent of i.c’s has not had much effect on the total engineering effort needed by the set maker.
Two or more circuits which perform properly on their own usually interact when harnessed together particularly if they are mounted in a compact assembly such as the G11. Furthermore, i.c’s owe their good performance to the use of sophisticated internal circuits which are easily upset by the presence of spurious inputs. Common examples are hum, field or line pulses on 1.t. or earth lines or picked up from electrostatic or magnetic fields generated by neighbouring conductors or components, or simply the all pervading fields from the scanning yoke and the line output transformer. For instance, a phase-locked loop in a line oscillator i.c. has high gain, and if any hum or field voltages appear in the loop the result will be curved verticals on the picture. Another example might be a high-gain clamp circuit in an RGBY matrix stage in an i.c. Any pick up of line pulses coinciding with the clamping interval will cause the signals to be clamped to the wrong potential.
A bit of thinking along these lines will make it easier to understand why it is that an important part of the development effort required to design a complete colour receiver consists of solving odd problems. This trouble shooting process can begin only when the first prototype receiver using printed circuit boards has been completed. Countless problems can be avoided by careful anticipation in the early stages of the design, but some unexpected ones always slip through the net. The only answer is plenty of hard work.
Prototype G11 receivers were subjected to a long series of field and home tests until every problem had been identified and cured. In most cases this involved modifying a printed board, and this took time. The following unavoidable sequence (for which there is no satisfactory short cut) had to be repeated: redesign of the board; preparation of artwork; cutting up boards; etching; drilling, and punching or filing specially shaped holes; procurement of components; assembly of new boards. Then the new version had to be retested under all the appropriate conditions to make sure that the cure really was effective.
At various stages in the development programme the design was temporarily frozen and new batches of prototype receivers were built, based on the most up-to-date specification. These of course had to be retested to make sure that all previous assessments were still valid, and that minor changes had not introduced new problems. For instance, re-routing a lead or conductor to prevent pick-up at one point might introduce it elsewhere. This kind of effect has to be guarded against as a matter of routine.
A stage was finally reached when repeated testing failed to show up any performance defects and the commercial department was satisfied. At this point the design could be said to comply with the electrical specification. It was not however the end of the development programme. A lot of work had been going on in parallel with the electrical design and more remained to be done.
Designing for Production
Design work for a chassis’ production has two aspects: electrical and mechanical. Continuing the theme that we have just been discussing, about electrical development, it can never be assumed that because a small batch of receivers works well the job is complete. Far from it. If mass production started on this basis the result would be certain and expensive catastrophe. The reason? Tolerances and also inadequate knowledge about what has actually been designed, quite apart from matters of safety and reliability.
The small mains panel is conveniently situated to enable the fuses to be quickly checked.
In the engineering division at Philips there is a separate laboratory completely independent of the design group. This checks and rechecks every aspect of the product. The laboratory makes comprehensive measurements of every part of the circuitry on complete batches of receivers. Any defects or doubts are referred back to the design team and appropriate action is then taken until the product is proved to be correct. Similar action is taken with regard to the mechanical aspects of the design. This activity is concerned with two main objectives.
First, it ensures that even when adverse component tolerances occur the receiver will still perform correctly this confirms the calculations and testing on which the design was based. Consequently a high degree of confidence is gradually built up so that when the receiver is mass produced in hundreds of thousands, every one leaving the production line will perform to specification.
The second aspect concerns production test gear, which has become highly sophisticated and in many cases automatic. Most of this equipment is specially designed for testing a particular assembly, say an i.f. board. It must not only be capable of measuring the performance characteristics of the i.f. response, it must also know what to accept and what to reject. It must be programmed to know the difference between a normal, acceptable variation in selectivity and an abnormal, unacceptable one. It can only do this if the correct assessment has been made of the fundamental design characteristics.
Test limits which are too tight result in rejection of perfectly good i.f. boards, whilst loose limits will result in boards which have a production defect resulting in poor picture quality being passed.
Efficient production demands good test gear programmed with the right information based on extensive laboratory measurements. Furthermore it is not just a question of setting limits. It is essential to decide upon the right combination of tests in order to achieve full control of what actually leaves the production line. All this work has to be completed well before production starts, and in the case of the G11 chassis all the necessary data was provided by the long programme of tests referred to earlier.
The other aspect of designing for production concerns the mechanical features of the product. The whole concept of the structural form of the G11 took careful account of the need to achieve a streamlined work flow in the factory. Every electrical assembly is pretested, and so is the cable harness. This means that at the casing stage where cabinet, chassis and boards come together, all units can be clipped or plugged into place with the minimum amount of time, handling and disturbance. Awkward operations such as hand soldering or putting in screws are avoided almost completely. In addition to being efficient in the production sense, it is also conducive to achieving the best possible quality. This is because any unnecessary handling of a fully tested assembly involves the risk of introducing a defect as a result of minor mechanical damage.
Subsequent operations consist of prolonged soak testing, routine adjustments of the preset controls and, of course, the purity and convergence, and final testing of every aspect of the receiver performance. When this testing is completed and the backplate is on, the receiver passes to yet another test station where a full customer check is carried out before it is packed.
It is not possible here to go into all the ways in which design techniques have been applied in order to make production more streamlined and consequently the output quality better. Numerous rules were followed in the design of the printed boards in order to make assembly and soldering easier and to avoid the risk of short-circuits or dry-joints; mechanical tolerances were thoroughly checked; many items involving assembly operations were designed so that special jigs and tools could be used to speed production, improve accuracy and reduce mechanical stress, and a number of operations were designed to be completely automatic. In short, the human element has been designed out as much as is practicable — because it’s fallible! It was considered better to demand skill in the engineering development programme rather than in the repetitive routine assembly process.
Aiming to produce a reliable product is not so much a matter of technical skill as an attitude of mind. If the need for reliability is considered at every stage of the design and production process the final product will be good. It is not sufficient merely to mount a reliability campaign after a new product has started production and the engineering pressures have eased off.
In the case of the G11 a high standard of reliability was required right from the start: this involved a long programme of testing and assessment running in parallel with the other development activities. Many design features contributed to what promises to be a good result. A few of them are listed below.
Careful and detailed checking has ensured that all components are operated well within their published ratings — in particular, large heat sinks have been used so that all semiconductor junction temperatures have a large margin of safety (look at the line output transistor heatsink). Circuits with low heat dissipation have been chosen in order to reduce the general operating temperatures, and these have been measured all over the chassis assembly and reduced still further whenever this has been practicable.
Many thousands of test flashes have shown that the pre-cautions against damage due to tube flashover are fully effective; special quality tube first anode potentiometers and focus units give freedom from drift of the grey-scale or the focused spot quality; the new interconnect system adopted was described earlier; many components have been subjected to endurance and type approval tests and some have been rejected as inadequate; and life tests of complete receivers have clocked up many tens of thousands of hours. And this is only part of the story.
The first life test was carried out on a large batch of early prototype receivers. Inevitably a few failed in the first thousand hours of cycled on/off operation at a high ambient temperature. The vitally important and encouraging factor however was that the failures were not random in nature but were due to only two simple causes. They were very carefully investigated, the underlying reasons understood and cured, and subsequent life tests showed no recurrence of the trouble. During recent tests the failure rate, even on hand built prototypes without the advantage of proper production test gear and tooling, was similar to the best obtained from an experienced production line and a well established product.
The auguries for the future look good therefore, and are backed by a large quality control department which monitors production quality on a continuous basis. Various techniques are used, but perhaps the most important is the permanent series of life tests that are carried on day and night all through the year on large batches of receivers.
The engineers carrying out the detailed testing procedures referred to earlier also had the responsibility of ensuring compliance with the requirements of the national safety specification BS415 15. This is a highly specialised type of work which needs considerable experience in order to understand fully the practical implications of some of the more subtle requirements of the specification. The sheer volume of work needed is formidable, but it is all part of the process of designing a new colour TV receiver.
Philips have built up their own standards of good practice over the years, and in a number of cases these are in addition to the official ones. The most important one insists that all material inside the cabinet must be flame retardant or non-flammable, whether it be an electronic component or a simple plastic clip. It costs money, but the improvement in overall safety is considerable. It will also be noticed that leads carrying high voltages have special heavy duty insulation, while those combining high voltages and currents in the line scanning circuits are duplicated in case one lead has a poor contact or gets broken. This avoids the possibility of a high energy arc occurring in a fault condition. In addition, special care is taken with the printed copper conductors carrying these high currents.
This brief description of some of the design philosophy and the procedures that gave birth to the G 11 has done little more than scratch the surface of a very large amount of engineering activity. Next time you have a chance to inspect a Gil chassis, check its performance, look at its styling presentation, and try to assess the engineering inside.
The Electrical Design Philosophy
The starting point was probably the decision to use the Mullard 110° 20AX colour display tube. This tube with the associated deflection yoke has an inherently self-converging action, only minor correction being needed to take manufacturing tolerances into account. This leads to major simplification in the circuitry and the preset adjustments required on the production line. A further advantage is that a very slim cabinet can be used, and this brings us to the next point. The levels of heat dissipation typical of older colour receiver designs would have caused rather high temperatures in the slim cabinets proposed by the stylists, a situation which would not have been compatible with the high degree of reliability that was required. It was considered important therefore to use more efficient circuits wherever a significant temperature reduction could be achieved.
The next factor was the decision to use well established circuits wherever practicable, and to avoid some of the new and exotic ones which present difficulties to the service engineer and show little benefit in terms of cost and performance. It is worth pointing out that design continuity contributes towards greater reliability, since the lessons of the past are the engineering capital of the future, and can be used to great advantage.
Each printed board had to be of a convenient size and as functionally complete and independent as possible. This simplifies all stages of assembly, alignment and testing, and provides built-in flexibility. It enables any assembly to be subsequently updated without destroying the principle of interchangeability of units for easy servicing. When new integrated circuits or design techniques come along they can be incorporated in the G11, if they confer substantial advantages, without undue disruption to service procedures and spares stocking policy.
There were other factors influencing the choice of circuits of course. All engineers, whether individually or as a team, have their preferences for particular circuit techniques, based on their own judgement and past experience. Also, Philips have very large international resources in terms of design and production of nearly all types of components —both active and passive. There are also applications and research laboratories that serve the whole company. Obviously, well established in-house techniques and components are often preferred to less well known ones from outside which have not been subjected to the same depth of investigation and testing.
Printed Board Assemblies
The problem here was first how to divide the circuits into functional units, and secondly how to divide the available printed board area into separate assemblies matched to circuit needs both in size and location within the cabinet. This jigsaw puzzle involved a lot of thought. Suffice it to say that in the end a satisfactory arrangement was achieved.
This is jumping ahead somewhat because a complete circuit diagram obviously had to be drawn up before this process could be carried out. It enables us to describe the contents of each printed board however whilst considering how each circuit came to be chosen.
The circuit has been divided up into the following units: power supply; line scanning; r.f., i.f. and sound; decoder; timebase; mains input; c.r.t. base; convergence tolerance correction; customer controls. We’ll describe each in turn, bearing in mind that it’s not possible here to give full details and circuit diagrams. It is hoped to provide a more complete description of certain key features at a later date.
The Power Supply
A stabilised power supply was considered essential for the G11, and design studies showed that a thyristor controlled circuit had advantages over other types available. It also made good use of a large amount of previous design experience.
A full-wave 100Hz circuit was adopted in order to avoid drawing d.c. from the mains supply, but it differs in some interesting respects from typical circuits of this kind. As mentioned earlier it was considered important to reduce heat dissipation, so it was decided to use active filtering and to omit the resistor which limits the current surge at switch on. Both these factors required the use of a “slow-start” circuit to reduce the size of the charging current pulses into the reservoir capacitor immediately after switch on to amplitudes comparable to those during normal operation. This had the advantage of reducing the electrical stress on several important components to levels well below those typical of many thyristor controlled circuits.
In order to ensure reliable operation of the slow-start circuit under all conditions it was necessary to adopt a two thyristor bridge configuration. The slow-start and voltage control circuit are of conventional type, but an inhibit circuit has been added. This positively prevents spurious triggering of the thyristors at wrong phases of the mains input waveform. Spurious triggering can be caused in a variety of ways, including very large voltage transient spikes on the mains supply exceeding 1,000V peak. The result can be that the reservoir capacitor is charged to the peak of the mains waveform at up to 370V instead of about 160V. This inevitably leads to over-stressing and the failure of several semiconductors. The inhibit circuit prevents this.
Early types of thyristor have been known to fail, going short-circuit with the result that the reservoir capacitor receives an excessive charge. This type of failure is almost unknown with the new types of thyristor used in the G11, but in compliance with the safety requirements of BS415 a crowbar circuit has been connected across the reservoir capacitor. It consists of a neon glow switch and a fusible resistor, and is an improvement on earlier versions of this technique because it is immune to short duration voltage over-loads and so cannot operate spuriously.
Again in compliance with BS415, a further safety circuit has been incorporated to inhibit the power supply and reduce the h.t. to a low level if a fault condition causes excessive e.h.t. beam current.
Although the basic mode of operation of this twin thyristor power supply is quite straightforward and should not pose any undue difficulties to service engineers, it is in fact the result of a very large amount of detailed design work. All sorts of electrical hazards such as c.r.t. flashover, voltage spikes on the mains input, fault conditions of all kinds, and unexpected circuit interactions have been investigated in depth and rendered harmless. It is hoped to describe some of this work in detail at a later date, as it constitutes a fascinating electronic detective story.
The r.f./i.f./sound board accepts the aerial input and provides the following outputs: chrominance subcarrier and luminance to the decoder; video to the sync separator; a.f.c. to the tuning circuits; and audio drive to the loudspeaker.
The tuner is the new Mullard U322. It combines a good noise factor — typically 7dB — with good signal handling capability. It should therefore give a good account of itself under noisy conditions in fringe areas whilst also having good immunity to cross-modulation when used close to a transmitter.
The tuner is followed by a block filter which provides most of the i.f. selectivity. Extensive use was made of computer aided design techniques and this helped to achieve a good group delay response. This means that all signal components, of whatever frequency, are delayed by approximately the same amount in their passage through the i.f. circuits. The amplitude response has been carefully matched to the vestigial sideband transmitter characteristics and the combination of these two factors, which are not easy to achieve together, has resulted in a good overall i.f. response and hence a high standard of picture quality. It is expected that engineers will notice an improvement compared with the G8.
The block filter is followed by three stages of wideband gain with a.g.c. applied to the middle stage. It is controlled by a potentiometer which is preset to transfer the a.g.c. action to the tuner at signal levels above 3mV. This gives the best compromise between noise performance and signal handling capability.
The i.f. gain section is followed by a TCA270 synchronous detector i.c. which also provides the a.f.c. and a.g.c. voltages. This i.c. was chosen for its good performance as a detector and for its high-gain a.f.c. action which maintains accurate r.f. tuning. Two video outputs are available. One feeds the sync separator on the timebase board whilst the other has the appropriate filters and transistor buffer stages to provide the 6MHz intercarrier sound, chrominance subcarrier and luminance signals.
An important design point is that all tuned circuits in the receiver — except for the quadrature coil in the f.m. sound detector circuit — are contained in screened units and are independent of the external circuit. It should not be necessary therefore to realign any circuit in the field unless the units themselves are being serviced. Replacement units can be fitted at any time without further adjustment.
The 6MHz intercarrier sound and audio output circuits are similar to those used in the G8. An improved version of the TBA750, the TB A750A, provides F.M. detection and D.C. volume control action. A discrete output stage using two BD131 transistors is retained but has been uprated to give approximately 3W output from a 25ohm loudspeaker. An i.c. output stage was investigated but found to give no overall advantage. The well established circuit was therefore preferred.
It will be noted that the luminance delay line is mounted on the i.f. board. This was done for two reasons. The delay time is a function of the i.f. bandpass characteristics (for a given chrominance bandwidth), so the strip can be updated at any time, and a new delay line used if necessary, without affecting the interchangeability of the boards—both old and new. The location of the line at the bottom edge of the board reduces the possibility of unwanted pickup from other circuits. A moulded cover has been fitted to prevent damage during handling.
A very stable 12V line is a great asset in signal circuits, particularly in the low-level decoder areas. It’s generated on the i.f. board by an overload protected high-gain TDA1412 i.c. stabiliser. This is supplied with a 17V feed from the line output transformer, and this voltage has been carefully chosen to cater for normal production tolerances whilst minimising the heat dissipated in the stabilisation process.
Engineers familiar with the G8 will recognise quite a lot of the circuitry mounted on this board. The same TBA560C and TBA540 i.c’s are used because, once again, there were no alternatives which offered sufficient advantages to offset the benefits conferred by familiarity in terms of circuit techniques, reliability, and ease of servicing.
The TBA560C is a luminance and chrominance control combination and is housed in a screened unit. This prevents unwanted interference currents from the all pervading magnetic fields generated by the deflection yoke and line output transformer being induced in the sensitive input circuits. It also saves considerable space and enables the complete decoder to be accommodated on a board of convenient size. The i.c. and its peripheral circuits provide D.C. control of brightness and contrast; it clamps the luminance signal to the correct black level and incorporates flyback blanking.
The chrominance signal passes through a bandpass filter to exclude luminance signal components and is then gain controlled to reduce unwanted changes in sub-carrier amplitude. This is followed by burst take-off and blanking, a D.C. saturation control, and sufficient gain to feed a discrete driver stage for the DL60 chrominance delay line.
The TBA540 receives the burst signal and with the aid of an a.f.c. loop and an external crystal generates the phase-locked reference carrier. It also provides colour-killer bias and ident outputs.
The output of the delay line has a very simple matrix circuit providing U and V inputs to the demodulators. At this point the circuit differs from that used in the G8 because synchronous demodulation and PAL switching are carried out in a TCA800 integrated circuit. This i.c. also matrixes the luminance signal with the three colour-difference signals to give low-level RGB outputs. These have a very accurately controlled black level, as a result of a high-gain clamp circuit in each of the three channels.
The TCA800 was chosen for the Gil because the peripheral component count is much smaller than in the equivalent circuits of the G8, giving a useful saving in complexity and space. It’s followed by three simple RGB output stages using BF458 transistors operating at relatively low gain. The simplicity of these circuits is in fact deceptive. The requirements are that the three RGB stages should track very accurately with each other over a long period of time, and that their frequency responses should be matched to within ±1dB over the whole passband.
Accurate tracking of the D.C. levels is achieved by mounting all three transistors on a common heatsink to achieve good thermal bonding. Thus differences in transistor junction temperatures are greatly reduced, and this prevents significant changes in the base-emitter voltages which would cause variations of the collector voltages and hence poor tracking between the three stages. Variation in collector load resistances would have the same effect; this difficulty has been overcome by using good quality power resistors which are conservatively rated and mounted upright on a common bracket to promote good cooling. Prolonged life tests on these output stages have shown that the stability and tracking performance are very good.
The physical construction of these circuits plays an important part in establishing the amount of stray capacitance present at the collector of each stage, and hence the frequency response. One of the last items in the design procedure was to match the frequency responses very carefully and to give them a high frequency roll-off to reduce the radiation of high order harmonics to other circuits.
Two i.c’s which are probably unfamiliar to many engineers are used, the TDA2590 and the TDA2600. The TDA2590 is a sync separator and line oscillator combination which can be regarded as a third generation version of a well established type.
The input consists of video from the TCA270 vision detector, and the sync is separated in a noise-gated circuit from which the field sync is immediately available as positive-going 11V pulses. The line sync is a fairly complex process however, resulting in accurate triggering of the line output transistor.
The line oscillator is synchronised via the usual phase-locked loop, incorporating a coincidence detector and flywheel filter together with an additional coincidence detector. Thus as an additional operation, the phase of the oscillator is compared directly with that of the incoming sync pulse and any error corrected. The oscillator drives a trigger pulse generator, and an output stage in the i.c. provides adequate base current for the usual line driver circuit. The finishing touch is provided by a line flyback pulse whose phase is compared with that of the oscillator: again any error is corrected.
The result of all this is that any change in the instant of the line output transistor’s switch off, caused for example by changes of loading such as a sudden variation of e.h.t. beam current, is corrected by reference back to the phase of the oscillator, this in turn being compared with the timing of the sync pulse. This process reduces line displacements caused by sudden changes of picture information. The phase of the picture relative to the raster can be adjusted by a preset potentiometer.
Additional features are the generation of an accurately timed burst gating pulse for the decoder, and a facility for D.C. switching the flywheel filter to provide a shorter time-constant for VCR operation.
The TDA2600 is a field oscillator and output i.c. synchronised directly by the output of the TDA2590. It was chosen for use in the G 11 because it operates in class D, which greatly reduces the power dissipation. In fact it is only about 4.2W for full scanning, and this can be handled comfortably by the i.c. and its associated heat sink.
The field oscillator’s sawtooth output is converted into a train of pulses which have a repetition rate of about 150kHz. They are of constant amplitude but vary in width. This in effect is pulse width modulation, the width of each pulse being a measure of the amplitude of the sawtooth at that particular instant. The train of pulses is used to drive the field output stage, so it will be clear that at any instant the output transistors in the i.c. are either turned off or are hard on, i.e. bottomed. This class D operation results in very low collector dissipation — only the switching loses. A low-pass filter after the i.c. integrates the current pulses to give the normal sawtooth scanning current required.
A sawtooth voltage proportional to this current is developed across R1 and is applied to a Miller integrator. The sawtooth drive to the base of the transistor produces a sawtooth collector current in the capacitor, and this provides positive feedback via Rs. The voltage across C is therefore the time integral of the current, and results in an almost pure parabolic output voltage at the collector. This waveform is amplified by driver and output stages and is used to load the diode modulator on the line scanning board in order to provide full EW correction of the raster shape and also control of picture width.
In these days of Teletext transmissions it is necessary to have very accurate control of field flyback blanking in order to ensure that the bright-up caused by the two lines of data transmission near the top of the picture is suppressed. The monostable circuit is used to generate an accurately timed blanking pulse and is triggered by the field flyback waveform.
In the early stages of the development programme some unconventional circuits combining the power supply and stabilised line scanning were considered. After lengthy investigations it was decided that these new circuits offered no overall advantages in performance, cost or simplification. Indeed, it was anticipated that the complex techniques used in such circuits would pose difficult problems to busy service engineers, who do not always have modern test equipment available, so the temptation to innovate was resisted. It was also reasoned that a large board carrying the whole power supply and line output stage would be too cumbersome and complicated, and in addition there was no sensible way of dividing it into two separate assemblies.
A circuit based on conventional techniques, using the new and much improved BU208A transistor, was adopted instead.
The line scanning board carries the whole of the line output stage and the associated circuits therefore, and is functionally complete and independent. The inputs consist of 156V h.t. and a low-level, synchronised drive pulse supplied to the base of an orthodox driver stage. This in turn drives the BU208A which operates as a switch in series with the line output transformer, connected across the stabilised h.t. line.
The basic mode of operation of the transformer will be familiar to engineers but it has two important and interesting features. First, it incorporates a diode-split overwind to generate the e.h.t. To summarise briefly, the overwind is a four layer winding, each layer being the full available width of the core window (allowing for insulation clearance). Each layer has its own diode rectifier, and the flyback pulse ripple is smoothed by the interlayer self-capacitance. The technology required to mass produce this overwind is impressive, even formidable, but extensive testing and field experience shows it to be more reliable than a tripler.
The other feature of the transformer is the use of a high-level diode modulator. The principle of operation is similar to that of the better known low-level type currently used by a number of set makers in 110° delta-gun c.r.t. chassis, but it’s simpler in configuration and does not need a large 1.t. load (or indeed any load at all) in order to keep the lower diode conducting. The sharing of the scan and flyback currents between the two diodes is a little complex, and a description is outside the scope of the present article. It’s hoped to return to this at a later date.
The diode modulator is driven by the parabolic waveform generated in the field timebase as described earlier. Thus the EW raster shape and amplitude are easily controllable by means of pre-set potentiometers.
Another interesting departure from previous practice is the absence of harmonic tuning. This is not practicable with an overwind having high self-capacitance, and is in any case rendered unnecessary by incorporating very tight coupling between the primary and the overwind. This is achieved by interconnecting a coupling winding placed under the overwind with one forming part of the primary.
L.T. supplies of 37V and 17V and a supplementary h.t. voltage for the RGB output stages are generated by normal scan rectification, in which the diodes conduct throughout the forward scan to give good voltage regulation. Boost h.t. for the c.r.t. first anodes is obtained by rectifying the flyback pulse at the collector of the BU208A.
The last important point is the focus voltage. This is conveniently obtained from the first layer of the diode-split overwind, at about 6.5kV, without further processing. The c.r.t. needs a d.c. focus voltage of about 4 .5kV, and this is produced by a thick-film potential divider specially designed to meet the needs of the G11. This unit ensures a very stable focus voltage, and the spindle projects through the board for easy access.
The CRT Board
This board is mounted on a new c.r.t. socket specially designed for the G 11 to suit the dimensional characteristics of the 20AX display tube. It has built-in sparkgaps for each pin except that connected to the focus electrode. Formed metal sparkgaps are entirely satisfactory at up to 2kV, but are prone to corona discharge at higher voltages. At best this results in a change of focus quality due to the voltage drop caused by the corona current flowing through the high source impedance. At the worst the corona causes sufficient ionisation for an arc to develop, with consequent risk of overheating and a possible fire hazard in addition to circuit damage. The focus sparkgap is therefore of printed/pierced construction on the printed board.
The other feature of interest here is that the three first anode potentiometers required for grey-scale tracking are mounted on this board for easy access and to save interconnecting leads. The potentiometers and their associated gun switches form a single unit in a plastics housing, complete with operating thumb wheels and levers. This unit has also been specially designed for the G11, to a very tight specification, to ensure that the voltages appearing at the sliders of the potentiometers not only have a very high degree of stability but also track together. Approximately 500V is applied across the potentiometers, and the outputs at the sliders track to within ± I V over several thousand hours of life testing. This helps to ensure good grey-scale tracking in the dark areas of the picture.
The term “specially designed” seems to have cropped up several times in these two articles. The fact of the matter is that whenever a component of suitable quality or function was not available on the open market, the production scale of the G 11 justified commissioning a new product. This was always procured based on a specification very carefully tailored to the needs of the circuit and the mechanical requirements.
Convergence Tolerance Correction
One glance at this board illustrates much more clearly than words do just how much complexity has been saved by the use of the 20AX system compared with its 110° delta gunned equivalents. The circuits are so easy to adjust that there is not much to he said.
Bearing in mind that the fundamental mis-convergence of the 20AX system is only a small fraction of that inherent in a delta gun tube, and that the tolerance correction circuits are operated solely by stabilised scanning currents, it’s unlikely that under normal circumstances the controls will need resetting in the field. They should be regarded as factory preset adjustments.
Purity, Convergence and Focus
Experience at the time of writing shows that the 20AX system is capable of very good purity and convergence and that these are easy to adjust. There are one or two hints and tips, however, which are not immediately obvious and which may be of help to engineers.
Take purity. The first step is to set up the picture properly and then get the convergence approximately correct. It does not have to be exact. Now switch off and leave the receiver to cool for a quarter of an hour. Step two consists of displaying a white raster, either from a pattern generator or by disconnecting the aerial and the luminance signal flying lead link on the decoder board. Turn off the blue and green guns by operating the gun switches on the c.r.t. board. Set the red raster to a medium’ low brightness level to avoid heating the shadowmask.
Step three — turn the yoke position lever to slide the yoke as far forward as possible. Then operate the two-pole purity magnets in conjunction with sliding the yoke backwards if necessary. A further small backward movement of the yoke, and possibly a touch on the purity magnet, will give perfect purity and the most central position of the raster. Keep the yoke as far forward as possible, because this ensures that the maximum margin is available to compensate for expansion of the shadowmask when it gets heated by high beam currents.
If you cannot get good purity with the yoke in its forward position, start again, but this time with the yoke as far back as possible. When you have achieved good purity, go on sliding the yoke forward until the purity just begins to deteriorate, then ease it back a fraction to get correct purity again. Always finish with the yoke as far forward as you can.
The next point concerns static convergence. This is adjusted by two multipole ring magnets on the tube neck —one for red/blue and the other for green on red/blue. Always adjust the static convergence as accurately as possible. The reason for this reminder is that any static error at the centre of the screen results in a corresponding error twice as large at the edges of the picture area. For best results the static convergence should be adjusted only after the receiver has been switched on for one-two hours at a fairly high beam current. This enables electrostatic charges in the area of the tube neck and gun assembly to become stabilised, and this condition will be rapidly re-established on each subsequent switch on.
Finally, focusing. This should always be adjusted at fairly high beam currents, and the object is to obtain optimum focusing of vertical lines about half way between the centre of the screen and the two outer edges of the picture, on the horizontal centre line. This gives best overall spot quality.