electronics j a R F circuit design Active aerial with SMDs Artificial intelligence Satellite TV receiving equipment Volume - 4 Number - 3 March 1986 electronics technology Electronically controlled cameras 3-17 Satellite TV receiving equipment 3-25 offering an alternative to BBC and ITV programmes The future for artificial intelligence 3-35 The accordion image sensor Designing a closed loudspeaker box a simplified approach to hi-fi quality sound reproduction projects DC operated battery charger how to charge NiCd batteries from a car battery Active aerial with SMDs .... MSX extensions tartridge board the concluding article deals w„h the colour extens.on board 3-48 “ “*■ “ ‘ " information Readers Services 3-64 FAST AUTOFOCUS BY MICROPROCESSOR Our cover this month shows the Minolta 9000 with in front of it the electronic circuits by which it is controlled Perhaps the outstanding feature is the electronic control of the focusing system, which is fast and accurate: a real aid to rapid photography. The next step in the development of cameras is replacing the film by a miniature floppy disc. guide lines Switchboard Classified ads Index of advertisers selex-10 Digi-course II (Chapter 4) .... Variable power supply Attachment for multimeters 3-65 3-74 3-74 3-51 3-53 3-56 THE PRECISION AND QUALITY you were looking for at realistic prices L'JYABASU PCB Drafting Aids now made in India. • Tapes in black, red & blue. All widths in inch or metric measure available ex-stock. • 1C Patterns and Donuts to be introduced soon. PRECIOUS ELECTRONICS CORPORATION 52-C, Chhotani Building Proctor Road, Grant Road (E). Bombay-400 007, Dealers wanted all over India Towards a common open network standard? More and more users — and prospective users — of communicating machines, irrespective of whether these are robots, computers, telephones, or a mixture of them, are envisaging networks that comprise machines of dif- ferent makes. There is a common belief that when this becomes possible, markets will expand more freely, because competitive power will then not be dependent on which particular manufacturer a dealer is tied to, but rather on the price function, and performance of the machines Most users are, understandably, in favour of a completely open network, ie, one that will allow any make of equipment contained in it to freely exchange information with any other make of terminal attached to it. There is a snag, however, or, rather, there may be. The successful interlinking of different makes of communicating machines requires an internationally accepted standard. The International Standards Organization — ISO — is developing a general data- communications standard called Open Standards Interconnections — OSI — which, it is hoped, will eventu- ally facilitate the linking of, for instance, computers from different manufacturers However, and here is the possible snag, IBM, which dominates the world market for mainframe computers (IBM and IBM-compatible computers account for over 80 per cent of the world market), has its own system for connecting in computers, called Systems Network Architecture, SNA. Some 20000 SNA networks are already fully operational. Competitors of IBM, fearing that the SNA standard may further increase IBM's share of the market (and thereby reduce theirs) are already cock-a-hoop with OSI, although this will not be fully defined for quite some time yet. Although IBM, like other industrial giants, is used to proprietary standards, which can be made to force users into buying only their products, it is carrying out research and development on OSI. In fact, last October it brought out a local area network — LAN — that is fully open to other makes of equipment. Moreover, spokesmen for IBM have on several occasions recently reiterated IBM's backing of OSI. Cited is, for example, the value-added network — VAN — that IBM will operate with Japan's NTT, and which will have to accommodate NTT’s open standard as well as SNA. At present, these developments look encouraging, and, sceptical though we may be, we must hope that the basis of a common interlinking standard will be agreed comprehensive & exhaustive document for amateurs and professionals./ mm Now You Can Learn All That You Want To Know About Computers Operation - Programming Hardware - Software The cost and complexity of home Computers is a serious deterrent to the newcomer to computer operating and programming. We know of many people who would like to build their own but who lack necessary technical knowledge. The Junior Computer has been designed (for just this reason) as an attempt to open the door' to those readers who need a push in the right direction. It should be emphasised that, although simple to construct, the Junior Computer is not a 'toy' but a fully workable computer system with tne capacity of future expansion. It has been designed for use by amateurs or experts. Published by ELEKTOR ELECTRONICS PVT. LTD., 52 - C Proctor Road, Bombay - 400 007 Phone : 36 74 59 / 36 94 78 S8S oou,< . c & OC ''jEfftO » mm ?^ WOm5t ° nTe ,t«< , wide aPP", c uacWf>e lo0 ' s S V °oo^r, Katara Mansion. Post Box 16551. Worli Naka, Bombay-400018. Ph: 493-9544/45.493-5565 Telex:1 1 -75217 Grams: AMPVOLT. TIME TESTED ELECTRONIC TIMER A new concept in Electronic Time Delay is | introduced. Our up-to-date and innovative technolosy sives trouble free service and long operational life. ADVANTAGES : 1 . High repeat accuracy±1 %. 2. DPDT relay with contact rating of 10A. 3. Two mode of operation-delay on energize/interval. 4. Fifteen voltage ranges from 12 VAC/DC to 240 VAC. 5. Three types of terminations. 6. Available upto 60 hrs. Time range. 7. More than 4000 types to choose. METAL FILM RESISTORS FOT Fa *5 Roloy , For further details contact : :OENIG ELECTRONICS & INDUSTRIES (P) LTD. 54, INDUSTRIAL AREA, PHASE-II, CHANDIGARH-160 002. Indian Engineering Company Delias Eleatrcnias Pis. lid. ItCJl Semiconductors KOIMTAKT Cleaners 'RAYTHEON Semiconductors Semiconductors Texas Instruments Semiconductors m p Trimmin8 Potentiometers elekt©f Magazines c IZUMIYA 1C INC. P.C.B. Drafting Aids TELEDYNE Ribbon Cable Connector Panels Semiconductors RS 232 Connector Panels SPECIAL EX STOCK OFFER D-RAM 4116 41256 4164 42128 Regulators 7805 - 24 7905 - 24 SCR/Triacs 4 amp - 40 amp 400 V - 600 V TTL/CMOS 74 LSOO - 629 4001 - 402571 E-Prom 2716 27128 2732 27256 2764 Bulk importers and users in India may Kindly forward their specific enquiries We stock: Device offers ELECTRONIC EDUCATIONAL & TESTING DEVICES 101 Kitchener Road *02-04, Singapore Electrical. Electronics and Hardware Centre, Singapore 0820. Telex : DEVICE RS 33250 FOR ENQUIRIES CALL: 298 6455 (4 lines) :3.13 electronically controlled cameras Ever smaller chips with ever larger capacities form an ideal partnership with the modem camera. We thought it would interest a great many readers to find out how the two fit together, and based this article on the Minolta 9000. When Daguerre laid the foundations of photogra- phy In 1839, he also started the development of fhe photographic camera. For more than a hundred years, the camera re- mained a purely mechan- ical device. From a cumbersome square box with a fixed lens, it slowly turned into a small and handy piece of precision engineering, that offered more and more facilities. It is only relatively recently that electronics began to be used in cameras. True, the exposure meter of thirty, forty years ago used a selenium cell and a very sensitive moving coil meter. For many years this type of exposure meter was the only photo- graphic aid that used electronics. Later there fol- lowed the CdS exposure meter, and by this time it had become small enough to be built into the camera. But true elec- tronic components were then —some twenty years ago— still too large to be fitted inside a camera. It was only when the tran- sistor became miniatur- ized and integrated circuits (chips) made their appearance about fifteen years ago that camera manufacturers began to see the advantages of complementing the con- ventional mechanical parts in a camera with electronic devices. And soon electronics proved to be not only cheaper In production, but also capable of giving more accurate and better reproducible results. Re- liability remained a weak point for a time, but not for long. The results were semi-automatic cameras: electronically controlled shutters; programmable automatics; and others. These developments gave rise to the modern camera in which virtually everything is controlled by electronics. Even focusing is now accomplished with the aid ot a small motor, so that the photographer can concentrate wholly on the subject and com- position. Such a camera Is, of course, an ingenious piece of engineering as may be gathered from the photograph on p 19. What facilities? After first looking at the facilities of fhe Minolta 9000, we will describe how all these are realized by electronics. ■ Automatic focusing system: when the shutter I release is half depress- ed, the subject is auto- I matically put in focus A memory makes if poss- ible to locus first and choose the subject afterwards. ■ Electronically controlled shutter with exposure times of 1/4000 to 30 s ■ Exposure meter with a choice between inte- gral and spot measure- ment. With spot measurement it is furthermore possible to measure the lightest and darkest part of the subject separately. Again, a memory is pro- vided for storing the measured values ■ Exposure modes: (a) manual: (b) aperture priority auto exposure; (c) shutter priority auto exposure; (d) program- i mable — in this mode I the camera itself selects the f-number and the shutter speed. ■ Through-the-lens flash measurement, enabling fhe use of all types of exposure automatics Red LEDs in the flash unit are activated auto- matically when the am- bient light is insufficient to allow fhe camera to be focused. ■ Advanced peripheral equipment, such as a flash unit with zoom reflector that automati- cally sets itself to the focal point ot the lens in use; motor drive with 5 pictures per second and autofocus priority; databack with multi- spot metering facility, in- terval timing, and a fa- cility for making I individual exposure pro- Fig. 1. Block schematic of a typical electronically con- trolled camera: here, the Minolta 9000. All control is vested in two micropro- cessors. Fig. 2. Construction of the photocell. The whole area of the cell is used for inte- gral measurements, but only the annular part at its centre for spot measurements. grammes; and a separate exposure meter that can wireless convey the metered in- formation back to the camera. On top of these there are some other noteworthy facilities. It is, for instance, no longer possible to set the f-number and shutter speed manually: the whole range of (-numbers and shutter speeds must be scanned with the aid of small slide switches until the correct values have been arrived at. Film sen- sitivity is set with a push- button — it can also be done by the film itself with the aid of the DX code printed on it. It takes some time, therefore, before you are used to this camera, because the usual rotary mechanical switches are conspicuous by their absence. The central processing system The monitoring and con- trol of all these facilities re- quire no (ewer than 150 000 transistors in the shape ot two micropro- cessors and some smaller ICs. The block diagram in Fig. 1 shows what is con- trolled by the two micro- processors. The central processor serves all general facilities, while the second deals exclusively with the autofocus. All other blocks within the dashed lines are separate ICs Outside the dashed lines are the operating switches and push-buttons; the control devices such as the magnetic switches and the autotocus motor; the displays; a charge- coupled device —CCD; encoders; and the various connections between the electronics and the peripheral units The central processing unit (CPU) receives a great number of inputs from various sources A pair of contacts in the camera teed information as to the sensitivity of the film used to an integrated circuit that decodes and memorizes the information in digital form. The film carries a so-called DX code for this purpose. The memory ot the 1C can be read at any moment by the CPU. Each autotocus lens con- tains a read-only memory (ROM) In which the prin- cipal data ot the lens are stored: smallest and largest aperture, and focal length. These 8-bit data are read by the CPU thirty times per second. This has been so arranged be- cause, when a zoom lens is used, the focal length changes every time the zoom Is adjusted. Slide contacts in the lens enable the code for the focal length to be con- stantly matched with the actual value. In this way, the CPU is ted with up-to- date lens information at all times In a zoom lens the ROM also arranges the conversion of the slide contact positions into a serial data stream. The CPU also needs the in- formation as to focal length for the autotocus processor and for the reflector position of a (lash unit. The connections to the motor drive and the camera back primarily use serial data streams also The CPU is connected via an interface to all parts that switch, monitor, or sense anything in the camera; to peripheral units such as a flash gun or an infra-red receiver; and to the exposure meter. The exposure meter con- sists of an integrated cir- cuit that evaluates the amount of incoming light with the aid of a photo- diode at the bottom of the mirror compartment and converts this analogue value into binary digits (=bits) that are fed to the CPU. The photodiode is a very fast type, because it not only serves to sense the amount ot ambient light, but alo that of flash light. The information as to flash light is. however, used in analogue form, because digitizing and processing it would take too long. Electronic flash units pro- vide flashes ot between 11000 and 1/50000 second. The photodiode measures the amount of flash light that falls onto the film, and as soon as this reaches the required value, it signals to the flash unit to stop the flash immediately. This clearly illustrates the necessity for a very fast photodiode The Minolta 9000 uses a ' very practical method of (electrically) switching be- tween integral and spot measurement — see Fig. 2. I This photograph shows the multitude of electronic devices: it is almost unimaginable that all this —and the many optical and mechanical parts— fits into such a small case. The photocell has the same length-to-width ratio as the window. In integral measurements the total amount of light tailing onto the cell is measured, whereas in spot measurements only the light falling onto the cell through the annular con- ductor is taken into account. The encoders connected to the interface 1C consist of tiny cog-wheels and opto-couplers. One of the cog-wheels is connected to the autotocus motor and the other to the f- number control. In this way, the CPU obtains infor- mation as to the angle of rotation of the autotocus drive motor and the f- number setting. The magnetic aperture switches ensure that the shutter is released at the right moment. The two magnetic shutter switches operate the first and second section of the metal shutter respectively: Fig. 3. The autofocus system: 3a shows how the beam of light travels from the object to the CCD; 3b shows how two identical images are projected onto the CCD element with the aid of two lenses; the distance between the two images and their location on the CCD give an indi- cation of the state of focus of the object (3c). the interval between the two operations is deter- mined by the CPU. The operating controls of the camera are shown at the left-hand side ot the block diagram. In reality, each of these is a simple push-button or slide switch. The two liquid-crystal displays (LCDs), one in the viewlinder and one at the top ot the body of the camera, are controlled by a separate 1C The displays give information as to shutter speed, f- number, the selected ex- posure programme, the method of measuring the exposure time, and any corrections. The exchange ot infor- mation between the CPU and the autofocus pro- cessor will be described later in this article. It is clear trom the descrip- tion so far that the CPU is the brain ot the camera that constantly receives, processes, and transmits data for the operation and control of the various parts of the camera. To this end, it contains 3 Kbyte of software (mask programmed ROM), and some 100 bytes of random- access memory RAM for temporary storage of data. A noteworthy aspect is the clock frequency which, at 4.2 MHz, is higher than customarily found in CMOS processors. The autofocus system The autofocus system con- sists of a microprocessor 1C a charge-coupled device (CCD), and a small but powerful motor. The processor, which has a 3 Kbyte programme, receives information from the CCD via an interface and on that basis, and in conjunction with the CPU, drives the motor via a separate driver 1C The CCD is an image sensor containing 128 sequential image dots. A tiny part of the centre of the field of view is pro- jected twice via two small lenses onto the series of dots, as illustrated in Fig. 3a. The image sensor is located at the bottom of the camera and obtains its information from an auxiliary mirror that is situated behind the main mirror and immediately in front of the shutter. This process is shown in slightly different form in Fig. 3b The double projection onto the series of dots is shown in Fig. 3c If the ob- ject is sharply focused, each image occupies a certain number of dots at a certain location on the CCD. All dots are continu- ously scanned by the in- terface 1C which converts the measured analogue value of incident light into binary data. This infor- mation allows the autofocus processor to j determine the exact lo- cation of the two images I on the CCD. When the ob- ject is not in focus, the two images will be further apart or closer together. The autofocus processor calculates the distance between the two images and from the result it can determine into which direction the lens must be turned to obtain a sharp focus. The perfection with which this happens is il- lustrated by the fact that the drive motor is slowed down when the object is almost in correct focus, and shorted out immedi- ately it is in sharp focus. The motor position is then immediately stored in the CPU. We know from our own experiences that this system works fast and reliably. The only drawback is that if fhe tiny part of the field of view is evenly coloured and lighted, this results in Insuf- ficient information for the autofocus processor to function correctly. But in such a situation it is quite I easy to point the camera at a somewhat more con- 3-20 etektor India march 1986 Fig. 4. Another schematic representation of the interplay of the mechanical and electronic parts in the autofocus system. frosting part of the object, memorize this information by half depressing the shutter release and then pointing the camera to the wonted part of the ob- ject again. Conversely, the autofocus system may be switched oft, and the focus set manually. An interesting featue of the autofocus system is that if the ambient light has a value of less than 3 (with 100 ISO film), some red light-emittting diodes (LEDs) with reflector in the associated flash unit are switched on by the camera for a few seconds. These LEDs project a tiny red spot a) the centre of the viewfinder image, i.e. . from where the CCD gets its information. The spot contains a grid that pro- vides an artificial contrast, so that the camera can be focused In complete darkness. Exposure modes The Minolta 9000 has four exposure modes: (a) manual; (b) shutter priority automatic exposure; (c) aperture priority auto- matic exposure; (d) pro- grammable When the programmable rletoor India inarch 1986 3.21 Malns-operated NiCd chargers are in plentiful supply, but a NiCd charger that operates from a car battery and enables fast charging is something special. The one described here can charge 9-, 12-, or 15-volt batteries. DC OPERATED BATTERY CHARGER Lowering the e.m.f. — electromotive force - of a car battery is easily done with the aid of a resistor, zener diode, or voltage regulator, but rais- ing it is rather more difficult. The method chosen here is the familiar one of voltage doubling. How this is done in this charger is illustrated in Fig. 1. In Fig. la, switch S connects the negative terminal of electrolytic ca- pacitor Ci to earth, so that both C3 and C< are charged to the (car bat- tery) supply voltage Ui>: Uo = UC4 = UD2 + UC3 = Ut>2 + Ub — UD1 =Ub (1) In Fig. lb, switch S connects the negative terminal of Ci to Ub, so that the output voltage, Uo, becomes: Uo=UC4=Ub + UC3-UD2 = 2Ub— Udz (2) When the switch is returned to earth as in la, the potential across C< re- mains at Ub, because Cj cannot dis- charge It is clear from this that Uo ( = Uci) will alternate between Ub and 2Ub— Udz. If the switching speed is high enough, the output voltage will approach 2Ub— Ud 2. Circuit description In practice the switching is carried out by a Darlington pair of tran- sistors: T. -T; and T.-T. in Fig. 2. These transistors are controlled by an integrated circuit Type LM3524. Two of its features make this device particularly suitable for the present application: the push-pull output stage, which can drive the switching transistors, and the error amplifier. The error amplifier controls the width of the pulses at the input of the push-pull driver stage on the basis of the error signal at the output of the charger. The larger the deviation of the output current from the wanted value, the shorter the switch-on time of the power transistors carrying the output current. The voltage doubling circuit consists 3-22 elekto of capacitors C3 and C* and diodes Di and Dj. These diodes are fast recovery power types in a TO-220 case, which is readily mounted onto a heat sink. An oscillator in the LM3524 generates a rectangular signal for the T-type bistable and the two NOR gates, and a sawtooth signal that is applied to the non-inverting input of a comparator. The frequency, fa, of the oscillator is fo = l/2nRsCi = 1/295 x 10 6 = 3400 Hz. A reference voltage of 2.5 V is pro- vided by divider Ri-R* and applied to the non-inverting input of the error amplifier. The inverting input of this stage is provided with information as to the level of the output voltage via divider R2-R3. The comparator here functions as a pulse-width modulator. Depending on the level of the error signal at its inverting input, and the level of the triangular signal at its non-inverting input, the comparator produces a rectangular signal with varying pulse-width at its output. This output constitutes the real control signal for the power transistors. To ensure synchronicity and a 180° phase shift, the comparator output is applied to the bases of the drive transistors via two NOR gates. Pulse-width control has the advantage that the average Fig. I. In a, both Ci and Ci are charged to Ut minus the small drop across the relevant diode , in b, the output voltage is the sum of the voltages across Ci and Ci minus the drop across Dt The switch is controlled by an oscillator, modu- lator, and regulator. Fig. 2. The cir- cuit of the bat- tery charger consists essen- tially of the con- trol. which is contained in one Type LM3524 in- tegrated circuit, power switching transistors Ti to T* and the voltage doubler comprising Dt Dt Ci and Ci >3.23 ICi = LM3524 Fig. 3. The whole of the battery charger, down to the heat sinks, is contained on this printed circuit load current remains substantially constant. The current limiter — CL — in the LM3524 is not used in this appli- cation. Da, are fitted on the printed circuit board shown in Fig. 3. If the board is fitted in a case, there should be suf- ficient space above electrolytic capacitors Ci and C« to ensure good ventilation. Once the board has been com- pleted. the open-circuit output voltage should be measured. This should be somewhat higher than 20 V. Note that a perfect voltage doubling, ie. from 12 V to 24 V, is not possible because of the saturation voltage of power switching tran- sistors Ta and Ta and the forward drop across the power diodes. Next, the behaviour of the circuit under load should be checked with reference to Fig. 4. Our laboratory prototype has an open-circuit output voltage of 20.2 V. Under normal load conditions, the output voltage re- mains substantially constant (+0.5 V) until the load current exceeds 3 A. example, NiCd cells are normally charged with a current, Ic, of 120 mA to 400 mA. If ten of these cells are charged in series, there will be a drop, Ua, of 15 V across them. A cur- rent limiting resistor, Ra, should then be used, whose value is calculated Fig. 4. The out- put current i* output voltage shows that the output voltage re- mains substan- tially constant for load currents up to 3 A. Construction and test All components, as well as the heat sinks of the switching transistors. T< to T< , and the power diodes, Di and The power, Pa, dissipated in Ra calculated from Sintered-plate cells are normally rated at 1.2 Ah, and may be fast- charged with a current of 2.5 A for thirty minutes. HS:GS Fast charging During fast charging, the charging current must, of course, be limited in accordance with the requirements of the cells or battery under charge. For In satellite television, pro- grammes are beamed up to a satellite from where they are retransmitted to serve an area (called foot- print) that is impossible to cover with a terrestrial aerial. The satellites used for this are geostationary, that is, they orbit at the same speed as the earth's rotational velocity. This makes it possible for a re- ceiving aerial (called dish) to be (irmly locked into position. Any dish within the footprint should receive good-quality sound and vision. There are several satellites dedicated to broad- casting programmes, and these are known as Direct Broadcast Satellites —DBS. Among these are the Rus- sian Gorizont satellites which send programmes across the world to official Soviet ex-patriot groups. Such satellites have very powerful transmitters, so that only small dishes are required to receive their signals. Whilst many European countries, including France, Federal Germany, and the Republic of Ireland, are planning to launch and build DBSs, British plans to establish a DBS have been aban- doned, at least for the time being, because of the enormous costs in- munications Satellite 1). Between them they broad- cast seventeen channels, most of them in English. Both Intelsat V and ECS-1 are communications satel- lites used primarily to route telephone calls across Europe and to the USA. The footprints of these satellites are shown in Fig. 1 and 2. The NESAT system from NEC Business Systems has been designed to plug into ex- isting TV sets to deliver multi-channel television to a variety of consumers. With this system, customers need not wait to be hard- wired to a cable network; nor do they have to wait tor DBSs to be launched. The NESAT system has several unique features that may place the equip- ment well ahead of the competition in the race to become the number 1 supplier of satellite TV re- ceiver systems designed specifically to meet the high standards demanded by the British and Fig. 1. Coverage area ("footprint") of European Communications Satellite 1. Fig. 2. Coverage area ("footprint") of Intelsat V. >3-25 Channels currently available (1) via Intelsat V: Premiere — which shows recent box office movies for about nine hours a day. Children's channel — with programmes aimed ex- clusively at young children and teenagers for eight hours every day. Screen sport — sports and leisure programmes for six hours every day. MirrorVision — movies and entertainment pro- grammes for nine hours every day. CNN — a 24-hour news channel. (2) via ESC-1: Music box — pop music programmes for 18 hours every day. Sky channel - general entertainment for 16 hours every day. TV-5 — programmes from national French language stations for 3 hours a day. New world channel — a diet of religious programmes for 1 to 6 hours a day. WorldNel — news and information programmes from the US Information Agency for about 1 to 2 hours every day. SAT1 — a publishers channel broadcasting about 10 hours a day. Te/eC/ub — broadcasts mainly films for about 8 hours a day. Fi/mNef ATN — mainly films and entertainment for about 9 hours every day. World Public News (WPN) - mainly US news material for about 9 hours every day. 3SAT — programmes from German language stations for about 6 hours every day. RAI — an Italian public service channel. Europe TV — (formerly Olympus TV): programmes from European Broadcasting Union — EBU— member stations for approximately 3 hours a day. RTI-Plus — general entertainment for live hours a day. With the exception of Sky Channel, all these chan- nels are at present clear, i.e„ they require no decoding system. European markets. The NESAT system com- prises three main compo- nents: dish, low-noise converter (LNC), and in- door unit (IDU) tuner. Plan- ning permission may be required for the erection of the dish in certain cir- cumstances at the present, but restrictions and regu- lations are likely to be relaxed in the near future Many of the current regu- lations covering broadcast were evolved some time ago, when the possibility of utilizing near-earth or- biting satellites was un- dreamt of. The low-noise converter has a low-noise amplifier that uses gallium-arsenide (GaAs) field-effect tran- sistors which reduce noise and thus increase picture quality. The IDU tuner enables the user to preset different parameters adopted in the ECS-1 and Intelsat V satellites for each chan- nel. Selecting channels is from then on simply a matter of pressing the ap- propriate button on the front of the tuner. The tuner is designed for use with any type ot television re- ceiver. The unique feature of the NESAT system is the facility for simultaneous reception of differently polarized signals. Channel operators use either X or Y polariz- ation. NESAT is the only system capable of receiv- ing signals with both types of polarization and pass- ing them on to the TV set via one cable. NEC Business Systems (Europe) Limited is the British subsidiary of the NEC Corporation, which is the recognized industry leader in a variety of high technology electronics sectors. It is one of the leading and one of the largest electronics manufacturers in the world, with 70 plants throughout the world, and more than 74 000 people Fig. 3. NEC's 1.8 m dish with two low-noise converters, which are stacked to enable simul- taneous reception of horizontally (X) and vertically (Y) polarized signals. Fig. 4. Close-up of two stacked low-noise converters, mounted onto the dish as shown in Fig. 3. Fig. 5. NEC’s IDU (indoor unit) tuner, designed to sit below or above the TV set, is slimmer than most video recorders. making and marketing i products that are sold in | 140 different countries. NEC's experience in high technology space related telecommuni- cations systems goes back some 15 years. Around the world, there are no (ewer than 6000 NEC-equipped microwave stations, while satellite communications earth stations (recog- nizable by their large dish aerials and which have become a symbol ot the 20th century) manufac- tured and supplied by NEC account lor almost 50 per cent of the world's total installations. NEC are also involved direct in manufacturing equipment for use In television transmitters, and the company has had some 15 years' experience in transponder design. It recently won a contract to supply transponders lor use in all the Intelsat VI series of communications satellites. Transponders are the devices on board satellites that receive signals from earth stations and relay them back to receiving systems. Satellite TV Antenna Systems claims to have achieved a breakthrough in satellite TV reception by cutting the cost of receiv- ing equipment by half, largely through technical development, and expects that private ownership of such equipment will consequently be en- larged. The company, known as SATVRN, is offering com- plete systems that retail at starting prices of less than £1000. The systems have three elements: the dish aerial, which receives signals from the satellite; the elec- tronics head unit, mounted on the dish, which amplifies the signals and converts them for TV reception; and the tuner, which is plugged into the TV set. Dishes 1.2 metres across supplied by SATVRN can be mounted in the garden of a small home, on an outside wall, or on the roof. The firm also offers satellite master aerial TV (SMATV) systems, which are com- mercial installations suitable for hotels, housing estates, and apartment blocks. Hotels using these systems can offer TV pro- grammes from their own countries to foreign guests, for instance. SATVRN has supplied equipment to the US Navy, the European Space Agency, and customers in western Europe. Yugoslavia, Israel, the Gulf States, and Canada. Another breakthrough in satellite TV receiving equipment occurred in the home of electronics engineer Mr Steve Webb of Swinton, near Malton. N. Yorks. His three children induced him to design a simple means of receiving information being broad- cast by spacecraft. Ac- cording to Mr Webb, "games are useful to help youngsters get interested in computers, but they can become a total misuse of the technology. My children got fed up play- ing space invaders, so we set about trying to com- municate with two British satellites to get Information and pictures". Using the know-how he had acquired in 10 years' work on satellite systems with two major UK space companies, he worked for fifteen months to produce a receiving system that converts satellite signals and decodes them via a computer onto a TV screen. "The first receiver I built for the children was crude," said Mr Webb. "So, I de- cided to develop a fully automatic model for any- one to use." The result is a fully automatic version called ASTRID, acronym for Automatic Satellite Tele- metry Receiver and Infor- mation Decoder. The total cost ot ASTRID and ac- cessories is £149. One of the biggest associations of computer users has described the device as an "outstanding product and a major breakthrough, bringing many exciting oppor- tunities to amateur scien- tists and radio amateurs". Mr Webb believes the device will particularly appeal to schools in a whole range of related subjects ranging from geography and maths to science and computer and radio technology. The research and develop- ment work was funded by the Micro Metalsmiths Microwave Company of Kirkbymoorside, N. Yorks, which Mr Webb joined last year. ASTRID is reported to be attracting worldwide in- terest following tests by science teachers through- out Britain, associations of computer users, and trade publications. NEC Business Systems (Europe) Limited 35 Oval Road London NW1 7EA Telephone: (01) 267 7000 Telex: 265151 Fax: (01) 267 1645/1611 Sa/e/l/le TV Antenna Systems Limited 10 Market Square Staines Middlesex TW18 4RH Telephone: (0784) 61234/52155 Telex: 877440 Fig. 6. NEC manufactures and installs almost 50 per cent of the world's satellite communications earth stations, such as the one shown here. Fig. 7. Typical transponder as supplied by NEC for use in the Intelsat series of satellites. ACTIVE AERIAL WITH SMDS Now that surface-mount devices — SMDs — are becoming available, many readers will, no doubt, want to gain practical experience with these new components. What better way to start than with this tiny active aerial? As slated in Surface-mount Technology (Elektor India. January 1986), all major semiconductor manufacturers are heavily engaged in the develop- ment and production of surface-mount compo- nents. These components are much smaller than conventional ones and have no or very short con- necting terminals, since they are intended to be soldered direct to the cop- per tracks of a circuit board. In general, these boards no longer have holes drilled in them, other than for fixing purposes. It should be noted that, although all major manufacturers have a good range of SMDs in production, these devices may not yet be available from all distributors and stockists. working with surface- mount devices. It has been designed as an add-on unit for a car aerial and for portable receivers where a 12 V supply is available. The aerials used with these receivers usually have a fairly high resistance, whereas the receiver input impedance is typically ot the order of 50 to 100 ohms. The resulting mismatch has a detrimen- tal effect on the noise fig- ure of the receiver. they cannot enter the re- ceiver via the supply line. The 560 pF capacitor isolates the receiver input circuits from the DC supply. Note that the MOSFET has a typical mutual conduc- tance of 20 mS, so that it performs best with output impedances greater than 50 ohms. As the medium- and long-wave input cir- cuits of car radios are nor- mally high impedance, the present circuit will work well on those wave- bands. FM receiver inputs are generally low- impedance. so that the circuit will not be so effec- tive on the VHF bands. tip may be made from a length ot SWG20 (1 mm dia) bare copper wire wound around the heating element of the iron. Useful tips on mount- ing the devices are given in Surface-mount Technology in the January 1986 issue of Elector India. The component layout is shown in Fig. 2. In portable radios it is advisable to solder the aerial termi- nation direct to C>. Note, however, that the present circuit can only be used if the portable radio has a separate aerial input that bypasses the built-in ferrite aerial. The present circuit pro- vides a large degree of correct impedance matching via a dual-gate MOSFET, Ti. The aerial signal is applied to gate 1 of the device, while the potential at gate 2 is ar- ranged at half the supply voltage, i.e., 4.5 to 6 volts. The MOSFET amplifier is coupled to the receiver in- put via a short length of screened 75-ohm cable (as normally used In car radios). The conductor in this cable also serves to connect the supply voltage to Ti. The chokes present a high im- pedance to frequencies in the receiver range so that Finally Construction Since it is impossible to achieve absolutely correct impedance matching, the cable between the pres- ent circuit and the re- ceiver may radiate. It the resulting signal is picked up by the aerial, the MOSFET stage may oscillate. All this can be prevented by winding the initial length of the con- necting cable around a ferrite toroid or rod as shown in Fig. 3. JB:BL Note that the circuit board is not available ready made through our Readers Services. It is best made from the pattern on page 44 or from a piece of prototyping board. Soldering should be car- ried out with an iron rated at no more than 18 watts and fitted. with a sub- miniature tip to prevent damage to the fragile surface-mount devices. The Circuit description The active aerial presented here is a very simple circuit, which is pri- marily intended as a practical introduction to arch 1986 3.29 Cartridge board with user-programmable EPROM EXTENSIONS -2 I Second in the series on home-made MSX add-on units, this article presents a cartridge extension board and full details on EPROM-stored programs. As evidenced by the first part in this series Elektor India February 1986), the cartridge slot available on MSX type computers may be used to effect connection of home-made ex- tensions like the Elektor universal I/O bus. Usually, commercially available car- tridges merely contain an (E)PROM to run a program (game, utility). It is, therefore, possible to construct a device that will hold user-pro- grammed EPROMs whilst retaining the possibility to insert existing car- tridges. Our design offers the follow- ing facilities. 1. Easy connection of further hardware-extensions, like the Elektor universal I/O bus. 2. The present board may be con- nected to the existing 50-way out- put port of such MSX computers as the Spectravideo type. 3. The board may be used as an angled cartridge adapter or a ver- satile IC socket to hold several types of user-programmable EPROMs with 2, 4, 8, 16, or 32 Kbytes capacity. 4. The board is useful for the connec- tion of a Yamaha synthesizer. 3-30 elektor i The MSX cartridge As shown above, the present car- tridge extension board is the sort of design that many users would un- doubtedly like to see: universal, ac- cessible for measurements and experiments and with the possibility to insert one's own EPROMs. How- ever, before this can all come true, some knowledge is required of the ‘cartridge conventions’ used in MSX BASIC. We shall, therefore, first examine a typical MSX start-up pro- cedure. After power-on, MSX BASIC always establishes the amount of RAM (Ran- dom Access Memory) between ad- dresses 8000 and FFFF, and activates the largest continuous area en- countered. Next, BASIC examines slot address range 4000. . . BFFF. Each slot occupies 16 Kbytes, div- ided in four pages. At the beginning of every page, a sequence of codes is read to identify the slot contents. The bytes which supply this infor- mation are located in a fixed order, as shown in Fig. 1. The function of each code is as follows: ID (identification): a two-byte code that indicates the presence of a car- tridge (E)PROM. In that case, BASIC reads 41 hex and 42hex (ASCII A and B), respectively at these locations. INIT (initialization): a vector (address pointer) for the initialization routine associated with the cartridge func- tion. In case this is not required, a default value 0000 is present at these locations. STATEMENT: a vector pointing to the cartridge statement-handler, if applicable. If not, a default 0000 is present. For further details on this vector, refer to the user manual sup- plied with the computer or the car- tridge. DEVICE: a vector pointing to the cartridge device-handler, if ap- plicable. If not, a default 0000 is pres- ent. Refer to computer manual for further details. TEXT: a vector pointing at the token- coded BASIC program text in the cartridge. This pointer is of great in- terest to users who want to put their own BASIC programs into EPROMs. All foregoing addresses are stored in the cartridge (E)PROM with their least significant byte (LSB) first, as is customary in Z80 machine language programming. Practical circuit Actually, the present design, as shown in Fig. 2, is not much of a cir- 2 cuit at all; it is rather a truly universal and user-friendly IC socket for the 27XX series of EPROMs, ranging from the well-known Type 2716 (2 Kbytes) to the giant Type 27256 (32 Kbytes). Note that EPROM manufacturers have generally agreed on using the last two or three digits of the type indication to state the memory capacity in kilobits. Div- ided by eight, this will give the number of programmable bytes (one byte equals eight bits). To accommodate every member of the 27XX family, the present exten- sion board has a number of jumpers, which will have to be installed or removed as follows: jumper A selects between Types 27128 and 27256 EPROMs and should be installed with the latter type in- serted. jumper B connects terminal 27 of a Type 27128 to +5V. Thus: jumper A for a 27256, jumper B for a 27128. jumper C connects Vcc terminal 24 of 24-pin Types 2716 and 2732 to + 5V. jumper D connects address line An to terminal 26 of 28-pin Types 27128 and 27256. For the 2764, jumper C must be installed (pin 26 to +5V, not both jumpers C and D). jumper E connects terminal 23 (28-pin types) or terminal 21 (2732) to An and must be installed for all EPROMs except Type 2716. jumper F connects V PP terminal 21 of a Type 2716 EPROM to +5V. jumpers G, H, and I connect the EPROM CE terminal (chip enable) to MSX signal C5T, CSS or C512 in that order, C5T being the ROM select signal valid for address range 4000... 7FFF, CSS for 8000. ..BFFF, and C5I2 for both ranges, i.e. 4000. BFFF. Up to and including a Type 27128 EPROM, either C5T or C52 is u sed; a Type 27256 requires the CS12 signal. Table 1 summarizes all available jumper configurations in order that any user can readily find and set the jumper combination as required for the EPROM in use. So far, only EPROMs have been men- tioned because these are most readily available and programmable. However, it will be evident that pin- compatible proprietary PROMs or ROMs will work just as well. If fitted in the MSX computer, the in- 1 Fig. 1 These codes at the beginning of every slot address-block form a software visiting card’ of the cartridge, for identification by MSX BASIC. Fig. 2 Practical circuit of the cartridge extension board. The jumpers are set to suit the type of EPROM used (2. . ,32 Kbyte) Fig. 3 Pin designations of the popular 27XX series of EPROMs, arranged in order of memory capacity. sert/remove protection circuitry will detect the connection between SW1 and SW2 as present on the extension Three connectors are provided on the board; Ki is simply the edge of the extension board with connecting copper tracks on both sides for in- sertion in the computer cartridge slot; PCB connector K: is a standard male 50-way type (2 rows of 25 pins); while Kj is a cartridge slot connector with 0.1 inch pitch contacts, just as the one inside the computer. Construction Track layout and component mount- ing plan of the cartridge extension board are shown in Fig. 4. The ready- made PCB is a moderately sized, through-plated type, available as usual through our Readers Services. The soldering islands and slot con- necting tracks have been pre-tinned to guarantee stable contacts. Use of a 28-way ZIF (zero insertion force) socket is highly recommended because sooner or later EPROMs will have to be taken out, erased with a UV source, programmed again, debugged, etc., and this perhaps several times. The cheaper types of IC socket will inevitably develop bad terminal contacts after pro- longed use. . . Applications Now that a neat, universal (E)PROM socket is available, frequently used programs may be stored in a dedicated EPROM, just as with com- mercially available cartridges, but a good deal cheaper. However, before user programs may be successfully stored in EPROM, the MSX BASIC program storage method needs to be unravelled. Note that the following description does not apply to machine-coded cartridge programs, since these re- quire a more elaborate vector system. For a BASIC program, then, the ID and TEXT vectors are essen- tial; they are located at XX00-XX01 and XX08— XX09 respectively (see Fig. 1). Because the first 16 bytes of the cartridge (E)PROM are reserved for program identification and system vectors, the token-coded BASIC program itself may be stored from location XXI 0 onwards. MSX BASIC programs are generally stored in memory from address 8000 onwards, so the value 80 may be read for XX from now on. At 8010 the CPU must invariably read byte 00 The next locations contain a so-called link address (two bytes) and a line number (also two bytes); 3-32 elektor indta march 1986 Listing I This memory dump program may be used to analyse larger BASIC programs as they reside in RAM; it provides hexadecimal presentation of any given memory area and may be put in EPROM to function as a stand-by utility C3=47|i;10V Semiconductors: ICl =2716,2732:2764; 27128:27266 or corre- sponding pin-compat- ible (PIROMs. >y (2x251 B connector ay (2x25) Fig . 4 This through-plated PCB is small but effective when it comes to plugging in existing cartridges, hardware extensions, or EPROMs holding user programs. DUMP 10 CLS 20 INPUT-starfiA 30 INPUT“end" sB 40 FOR C = A TO B 50 LPRINTUSING“\ \* tHEX$( C )t tLPRINT” ” B0 FOR 0=0 TO 15 70 LPRINTUSIN6”\\" tHEX$( PEEK( C+D ) )i tLPRINT" 80 NEXT 90 C=C+!S:LPRINT" "tLPRINT" " 100 NEXT 1 10 END next comes a token-coded line of BASIC text, terminated with a byte 00 This procedure is repeated for the following text lines. To find out the hexadecimal codes that constitute a program, it is necessary to run the DUMP program of Listing 1, preferably with a printer connected to the computer. In case a printer is not readily available, the bytes may be put on the screen by changing all LPRINT commands into PRINT and next changing value IS into 7 in lines 60 and 90 to allow for the reduced number of printable characters per line. Note that the DUMP program may be 'attached' to any user program in memory by entering it from, say, line 10000 onwards. After RUN 10000 the program prompts for a start and end-of- program address; the former is always &H8000 the latter depends on the actual size of the program, which Table 1 Summary of the necessary Jumper configurations for every type of EPROM in the 27XX series. The choice between jumpers H and I depends on the selected memory area (see text). Table 2 This table is a hexa- decimal dump of the DUMP program as it resides in MSX computer RAM memory. All bytes have been analysed, and it may be useful to reconstruct program Listing 1 from it! 0 90 E4 22 5C 5C 22 3B FF 9B 28 FF j 97 28 43 79 Tk Tk • \ \ " ; Tk Tk ( Tk Tk : t C 44 29 29 3B 3A'9D 22 20 22 3B 0 81 I 80 50 0 PI ) , Tk " sp " EOL : L8981 "89 3.33 Table 3 These data are burned into an EPROM to function as a utility cartridge called DUMP. Compare the shaded addresses with those in Table 2 to note the move up by 10hex and the correspond- ingly adapted LSBs. 0123456789ABCDEF 8000 41 42 0 0 0 0 0 0 10 80 0 0 0 0 0 0 1 8010 0 17 80 A 0 9F 0 26 80 14 0 85 22 73 74 61 8020 72 74 22 3B 41 0 33 80 IE" 0 85 22 65 6E 64 22 8030 3B 42 0 43 80 28 0 82 20 43 20 EF 20 41 20 D9 8040 20 42 0 5E 80 32 0 9D E4 22 5C 20 20 5C 22 3B j 8050 FF 9B 28 43 29 3B 3A 90 22 20 20 22 3B 0 6D 80 8060 3C 0 82 20 44 EF 11 20 09 20 F F 0 8B 80 46 8070 0 90 E4 22 5C 5C 22 3B FF 9B 28 FF 97 28 43 FI 8080 44 29 29 3B 3A 90 22 20 22 3B 0 91 80 50 0 83 8090 0 A6 80 5A 0 43 EF 43 FI F F 3A 90 22 20 22 8OA0 3A 90 22 20 22 0 AC 80 64 0 83 0 B2 80 6E 0 | 80B0 81OOOOOOOOOOOOQO0| Fig. 5 The Spectravideo MSX computer may be connec- ted to the cartridge extension board with a short length of 50-way ribbon cable and two suitable sockets. is lengthened by some 160 bytes I because of the addition of DUMP. After this first aquaintance with the hexadecimal dumping format and use of DUMP in practice, the com- puter memory may be cleared (NEW) and DUMP entered as shown in Listing 1, i.e. from line 10 onwards. Run DUMP, enter &H8000 as the stan address and &H8100 as the end, and have a look at the machine code that constitutes this little program. With the use of Table 2, try to retrace the familiar BASIC lines to understand the MSX memory storage principle. Note that the link addresses and line numbers are in reverse order, that is with their LSBs first. All standard BASIC commands have a corre- sponding token-byte, and it will not be difficult to spot some of them: 82h=FOR; 9Dh = LPRINT; EFh = " = " (equal sign); 83h=NEXT; Flh=” + "; E4h= USING; etc. If this is all sufficiently clear, we will now consider the EPROM data. EPROM data It will be evident that the computer does not consider the machine code currently present in locations 8000 and up as located in a cartridge, because the identification group of bytes as already discussed is not present at the beginning of the program (8000. . . 800F). To obtain fac- tual EPROM data, the whole machine code program will have to be moved up by sixteen (10h) bytes, the link the identifiers placed at the begin- ning as outlined above. A practical example of how this may be accomplished is shown in Table 3; this is the DUMP program again, but this time as present in an EPROM; compare the data with those of Table 2 to gain an insight into car 1 ridge EPROM operation with MSX BASIC; program an EPROM with these data, plug it into the cartridge Z1F socket, and run your own utility cartridge. Finally, a word about lengthier, more complicated BASIC programs and their storage in EPROM. As already suggested, the DUMP program may be attached to them at a suitable high line number, e.g. 10000 With the main program fully debugged and operational, run DUMP, spot the link addresses, add 10ho« to them, move the program up by 10hex addresses, and write a suitable sequence of identification bytes. The link addresses always point to the next one, and are thus easily picked out for modification. Program end is marked by a link address reading 0000 but the real end, that is without the added DUMP program, may be found by looking for the hexadeci- mal equivalent of 10000 bytes 1027 in that order, next, change the preceding link address into 0000 Finally, note that programs run from cartridge may, of course, not be edited because they reside in read- only memory. Spectravideo connection The extension board need not always be inserted into the com- puter's cartridge slot; the Spec- travideo MSX computer, for instance, features a 'real' 50-way expansion connector for receiving an appropriate flat ribbon type socket. The present extension board is then connected with a short length of 50-way flat ribbon cable with such a socket on either end of it, as shown in Fig. 5. Note that there is a slight oddity with the Spectravideo output expansion connector; the tiny arrow on it does not indicate pin 1 as usual practice, but pin 50. However, no problems should be encountered if the example given by Fig. 5 is followed. This finishes the present article on MSX extensions; a further instalment will deal with the construction of a bus-board for this type of computer. GD:BL THE FUTURE FOR ARTIFICIAL INTELLIGENCE by Professor Margaret A. Boden, MA(Cantab), PhD(Havard), FBA Despite its short history, ar- tificial intelligence already promises to j change everyday life as much as the Industrial Rev- olution did. Machine intelligence was foreseen in the 19th cen- tury by Charles Babbage, whose cogs-and-gears calculating machine I worked in a way basically similar to today's com- j puters. A century later, I Alan Turing provided a theory about what ques- I tions could in principle be | answered by such a | machine. Artificial intelli- gence grew out of the work on digital computers in World War II, and was given the dignity of a name in 1956. Since the early efforts in the mid-1950s, it has had some notable successes. Today's computers can perform some of the tasks normally done only by our minds — though only to a very limited degree. For in- stance, some programs can respond sensibly to queries or statements ex- pressed in natural languages such as German or English — which means that or- dinary people do not need to learn a special programming language before they can interact with them. Expert systems Conversations with most of these programs have to take place over a teletype, but some can recognize spoken words. Other programs can de- scribe the shape and pos- ition of visible objects, and identify what they are. Still others can play games, or comment on events from a particular political stand- point. 1 And some can solve problems of various kinds, like those which an in- telligent robot would have to tackle The most publicly visible applications so far are the programs called expert systems. Some are already being used experimentally to give advice on medical diagnosis and prescrip- tion, genetic engineering, chemical analysis, and geological prospecting tor minerals and oil. Future expert systems will be used by ordinary families for example, to help motorists diagnose and fix faults in their cars. An expert system has built into it some of the theoretical knowledge and rules of thumb used by human experts. And it can be improved, up to a point, by adding new in- formation. So that it can help on a particular prob- lem, it is given the evidence that its human user has — it can suggest that relevant tests be done, if they have not been done already. Then it sup- plies an opinion based on this evidence To make it easier for people to evalu- ate its advice, the expert system can display its chain of reasoning. Current expert systems are very limited in what they can do, however, largely because they cannot reason about their own reasoning, or the user's reasoning either. They cannot explain their con- clusions differently to dif- ferent people, since they have no user-model in terms of which to adjust their explanations to a person’s level of knowlegde. But despite their limitations, a few cur- rent systems give more reliable advice than all but the very best human experts, and one or two surpass us all. The world expert on soya bean diseases, for example, is not a person but a program. Long term funding Government money from the western industrialized nations is being poured into artificial intelligence research, in both academic and industrial contexts The European Community has estab- lished the ESPRIT project for funding co-operation between its member countries in research into micro electronics and soft- ware technology. The first phase of ESPRIT will draw on £465 million from Com- munity revenues The British Government, as well as having a stake in ESPRIT, has set up the national Alvey Committee to recommed a strategy for the long term funding of artificial intelligence and related computa- tional techniques Govern- ment funds of £215 million have been allocated for this information technology work. The electronics industry is taking this research seriously too, matching ESPRIT'S £465 million with an equal contribution. And the Government's £215 million is also equalled by industrial money set aside for the Alvey research and development projects What are these machines of the Future the so-called fifth generation com- puters? The first four gener- ations are defined in hardware terms: machines based on valves tran- sistors silicon chips and very large scale inte- gration (VLSI). The predicted fifth generation is defined in terms not only of improved — massively parallel — hard- ware, but also of artificial intelligence. Multi-lingual robots It is hoped, for instance, to achieve reliable machine translation between various natural languages — even on texts that are not restricted to highly specialist subject matter. And some people forecast that computers of the 1990s will be able to inter- pret the speech of many different individuals, to act as intelligent assistants in a wide variety of tasks, and to provide advanced problem solving and sensori-motor abilities for mobile domestic and in- dustrial robots. However, achieving fifth generation computers will be much more difficult than most people assume. Once they have accepted the fact that some sort of machine intelligence may be possible, most people grossly underestimate the difficulties involved. One of the prime lessons of ar- tificial intelligence is the previously unrecognized richness and subtlety of human common sense, and the extent to which it guides our thinking. Never- theless, by 1990 the western nations will have a wide variety of commer- cially useful applications. It is not inconceivable then that artificial intelli- gence programs will be used by the general public at home. What is more, they will be used by many professionals whose decisions affect people's personal lives. Are there dangers in this prospect, and, if so, can they be avoided? The man in the street may place too much trust even in today's limited pro- grams. One might almost say that it is the unintelligent tasks that such programs cannot handle. Many of the things that our minds enable us to do are not normally called intelligent, because we can all do them so easily. Everyday abilities like talk- ing, seeing, or realizing friendliness from a facial expression, do not nor- mally need conscious ef- fort. Nor can we say how we do them. But they are far from simple. Indeed, their complexity — and subtlety — was not ap- preciated until researchers tried to model them on computers. The activities that involve our common sense pres- ent a daunting challenge to artificial intelligence. By contrast, many of the tasks we pay highly specialized professionals to do have proved more tractable. Ex- pert systems can already advise us on some of the recondite problems faced by mathematicians, doc- tors, genetic engineers, geologists, and chemists. This seeming paradox has surprised researchers, and most members of the public still are not aware of it. The widespread ig- norance of this fact can be dangerous, to some degree, for it means that most people lack a reliable sense of which questions current artificial intelligence systems can be expected to handle. It will be difficult to warn people of these dif- ferences as future systems achieve more humanlike programs whose limi- tions will not be so readily apparent to someone in- teracting with them. There is no hope of developing fifth gener- ation systems unless they can be made more human than today's pro- grams. This will require some basic theoretical advances. How smart are they? Tomorrow’s computers will need a better grasp of natural language, for example, and a better ap- proximation to common sense thinking. Without natural language they would be useless to the man in the street, who does not want to learn a special programming language, and they would be unable to interpret writ- ten texts or reasonably normal conversation. And without something like common sense, they would fall into all manner of absurdities. A future expert system could appear to have a fairly subtle command of natural language within the subject for which it was designed. Many users might therefore assume that it has a complete command of that language, at least in that subject. Some might even believe it to have a rich command of language in other areas too. These false assumptions could lead to its judgments be- ing given more credit than they are worth. Suppose the computer uses a familiar English word such as possible. The user knows that this word is similar in meaning to a number of others (such as probable, likely, con- ceivable. and so on), but also knows that it is not precisely equivalent to any of those, for each word has subtly different shades of meaning. Therefore, we should not assume that the words used by the computer, however well chosen in context they appear to be, have been carefully selected in preferance to other words carrying rather different im- plications. What of common sense? This is needed, for example, when someone has to make guesses about relevant facts. If one of these guesses is incor- rect, that new information can be used from then on. People can cope with the fact that a statement justifiably assumed to be true at one time can later be found to be false. Understanding limitations This cannot happen in tra- ditional logic, wherein truths are proved once and for all. And traditional artificial intelligence pro- grams are based on this type of logic Conse- quently, much research at present is trying to for- malize non-monotonic reasoning, in which truth values can shift from time to time as relevant infor- mation reaches the system. The limitations of artificial intelligence programs as well as their potential must be understood. In par- ticular, it must be realized that every program can in principle be questioned. The reason for this may be surprising. Programs are not objective systems that guarantee the truth, but rather subjective ones that represent the world in ways that may or may not be wholly veridical or reliable. An artificial intelligence program uses some representation of data, which may be partially false andbr incomplete. It uses rules of inference, which may be faulty in various ways — many will be hunches that are sen- sible only in certain cir- cumstances. And it employs decision criteria, or values, to select one course of action rather than another and these are essentially prob- lematic The crucial point, then, is that a program's data, inferences, and values can always in prin- ciple be challenged, just as they can when contain- ed in a human mind. Teaching work Some work has already been done on developing teaching systems capable | of encouraging this sort of computer literacy. One is the POPLOG system developed at the Univer- sity of Sussex over the past ten years for teaching arts and humanities students the principles of artificial intelligence programming. It is a user-friendly, interac- tive programming environ- ment, with a large library of "teach" and "help" files that enables students to learn at their own speed and in their own way. It is also a powerful research tool, since it allows the user to write programs in LISP, PROLOG, and POP-II. It has been recommended by Britain's central research councils as a main tool for current ar- tificial intelligence research. A system like this can be used to show students fairly quickly that an ap- parently intelligent program is neither so in- telligent as it seems, nor unalterable. For PROLOG helps the student to ex- plore and alter mini- versions of programs. Take ELIZA, for example, a relatively simple program that interacts with its user by way of English sentences. If you type into ELIZA the sentence My father drove me here, the program will answer: Tell me more about your fam- ily. or perhaps: How do you feel about your father? If you type in: / mistrust you. ELIZA responds with: Why do you mistrust me? This seems eerily humanlike. But if you were to type in: / b/gg/skxz you. ELIZA will just as hap- pily ask: Why do you b/ggtskxz me? In short, the program has no under- standing of English. It con- sists merely of a few simple rules of recognizing a few specific patterns or keywords and responding blindly to them in stereotype ways. Social im- plications No one knows what the ef- fects of artificial intelli- gence will be on our daily life in the future. Even the long term influence on unemployment is unclear. Some economists see a post-industrial society based on information technology, in which only a few people at opposite ends of the educational spectrum have jobs as we understand the term to- day, while others forecast a return to full employ- ment after the transition stage. All agree, however, that the pattern of employment will change, and these changes could lead to a more humane society. Many jobs will be available in the service and caring sectors of the economy: education, welfare and health care, entertainment, sport, and craft activities. These ways of spending one’s time are intrinsically more meaningful and satisfying than many — perhaps most — jobs in in- dustrial societies today. Moreover, even with full employment, working hours will probably be re- duced. So people will have more time to spend with their families and their friends than they do Finally, what ot the threat that artificial intelligence must deny our individu- ality and freedom? Many people fear this dehumanizing influence of computer technology. They would be right to do so it it were true that ar- tificial intelligence can allow no room for these aspects of human psychology. It is not. Artificial intelligence does not lead to a reductionist psychology — such as behaviourism was, tor example. Indeed, it is largely due to the in- fluence of computer based ideas that theoretical psychology now takes the mind, and mental representations or ideas, seriously again. Moreover, research in ar- tificial intelligence en- courages a new respect for the richness and power of everyday human men- tal processes. Artificial in- telligence can counteract the subtly dehumanizing influence of the natural sciences, in which there is no room for concepts like belief, reason, inference, purpose, and choice. THE ACCORDION IMAGE SENSOR Scientists at the Philips Research Laboratories have made a new type of solid-state image sensor. The new sensor has twice as many light-sensitive elements per unit area as previous sensors. This has been achieved without the need of a finer pattern for the electrodes applied to the sensor surface by tC technologies. The improvement is achieved by a new method of distributing the potentials over the electrodes, in this method a row of picture elements (pixels) is located under every two electrodes whereas four electrodes were previously required for each row. The availability of only two electrodes per picture element makes the transfer of the image information from the 'camera' section to the ‘memory' section (frame transfer ') rather more complicated. The potential hi/is that separate the information coming from the different individual elements are now stretched out one by one and then compressed again, tike the bellows of an accordion. In a solid-state image sensor, and also in a CCD (charge-coupled device) shift register, narrow paral- lel channels of /7-type material are located in a layer of p- type silicon. On the surface there are lin- ear electrodes, which are perpendicular to these channels. The electrodes are insulated from each other and from the silicon surface. It the silicon sur- face is exposed to light — through the electrodes — electrons are released in the silicon. If suitable potentials are applied to the electrodes (Fig. la), these electrons will build up charge packets under the positive electrodes in the n channels. In this way, charge is collected during a scanning period, with the size of the charge packets providing a measure of the local il- luminance in the image. Next, during the read-out phase, the electrode potentials are varied in such a way that the potential hills and valleys execute a 'peristaltic' motion (Fig. 1b), which transfers the charge packets from the image section to a storage sec- tion. From there they are read out line by line so as to supply the video signal. During the following scan- the top is a cross-section of electrode structure of a image sensor, in the longitudinal direction through an n-type silicon channel. One cell covers four electrode widths. Below the cross-section, the potential distribution during the recording of a picture is shown. The charge packets in the potential wells are indicated schematically, b) Sequence of potential distributions for transferr- ing the image information (to the right). Fig. 2. Schematic represen- tation of the potential distribution in an accordion image sensor at successive moments during the trans- fer of the image infor- mation from the image section to the storage sec- tion. At top left the first in- formation leaves the image section. The picture elements that initially cover two electrode widths are stretched out one by one over four electrodes: the accordion is ‘pulled open'. At bottom right the first in- formation arrives at the far end of the storage section. The information for one im- age point is once again ac- commodated in a storage element two electrodes wide: the accordion is sqeezed shut again. Fig. 3. The accordion im- age sensor The image sec- tion (dark) and the storage section (light) are at the centre. The electronic cir- cuitry for generating the electrode voltages is shown along the edges. Inset: enlarged view at the transi- tion from image section to storage section. Fig. 3 shows the complete image sensor described here, with part of the pic- ture enlarged. The results described here refer purely to laboratory research; they in no way imply the manufacturing or marketing of new products. ning period, the potential pattern on the electrodes in the image section is shifted by two electrode widths to give the usual television interlacing. Although three electrodes per cell would be suf- ficient for transfer of the collected charge, four electrodes per cell are generally used, as indi- cated in Fig. 1 This pro- vides simpler control and correct interlacing. If the conventional 3.5-nm technology is used in making a sensor, the cell dimensions will be fairly large. They can be made a little smaller by using a three-layer electrode struc- ture. If this is done, how- ever, the light incident on the sensor at some places must pass through three layers of electrode material (polysilicon) before it is detected. This gives a reduced sensitivity, particularly in the blue part of the spectrum. The accordion principle Two electrodes per cell are in principle sufficient for collecting the charge With this arrangement, however, charge transfer is not as simple as before, so the following technique has been devised. Instead of transferring all of the image information to the storage section at the same time, each charge packet is temporarily spread out in the space beneath two electrodes, and separated by a potential barrier two elec- trodes wide, beginning at the bottom edge of the image section. The con- ventional method of charge transfer can then be used, and the image information is 'peeled off line by line. The temporary 'stretching out' of the infor- mation disappears again when the charge packets reach the bottom edge of the storage section, so that in the storage section a row of picture elements again comes beneath two electrodes. All this is 3-38 elekto, indla march 1986 production method, the 3.5-> and Dz These variable- capacitance diodes are controlled by Pi: a high voltage across them causes a small capacitance, and vice versa. The frequency of an LC os- cillator is given by /= \HnTEc [Hz] (1) where /is the frequency of the oscil- lator, L is the inductance in henries (H), and C is the total capacitance of the two varactors in series in farads (F). The ratio between the lowest and the highest oscillator frequency, f> and fr respectively, depends on the square root of the ratio between the maxi- mum and minimum capacitance, Cz and Ci respectively, of the varactors: f'/r^ Cr.C, (2) The maximum capacitance of the Type BB106 varactor is about five times the minimum capacitance for a reverse bias voltage of 3 of 25 V, so that the frequency ratio is roughly 2.236, or rather more than an octave. The highest attainable frequency is around 300 MHz, but this depends, of course, also on the value of Li, The series combination L2-L1-L. is in- tended as a sort of wide-band choke. The inductance of Li (100 mH) is rather too large for high frequencies, because the reactance at those fre- quencies amounts to a few kiloohms owing to parasitic capacitances. Lower inductances are, therefore, used for the higher frequencies: La and L2. Inductor L2 is only of use at frequencies above 50 MHz: if the os- cillator is not required to work on these frequencies, this coil may be omitted and replaced by a wire link. BF900; BF905; BF907; BF961; BF981 BF494 ;3.49 Fig. 5. Suggested component layout of the RF test oscillator. Part9 list Ri;Ri;Ri = IOOk Ft? -470k R* = 330k Rs = 680Q R«=10M R7 = 220Q Pi = 50k linear preset Capacitors: Cr;Ct;C«=560p C. 68p Cr = 330p C.;C.= 1p Ci = I0p Cs:Cii=47n C>o = 100n b = 5 turns 0.3 mm dia. 130 SWGI enamelled Ls = ImH choke U = 100mH choke Semiconductors: Ti - BF900 or BF905 or BF907 or BF961 or BF981 Tr = BF494 Ti = BF246C D.;Dr-B8106 (see text) Ds = 1N4148 Type 85000 Fig. 6. Circuit of a possible 50-ohm one-step attenuator. The resistor values in the ac- companying table are calculated: in a practical circuit, the nearest stan- dard values should be used. Frequency range If varactors Type BB106 are used, the oscillator can be tuned over a fre- quency range of one octave, ie., the maximum frequency is about twice the minimum frequency. To cover a frequency range of, say, 2 MHz to 32 MHz (four octaves) four different coils are required for the Li position. Since it is not really possible to use a large tapped coil and a range switch — because the resulting stray capacitances would cause unreliable and unstable operation - separate plug-in coils must be used for In. At the highest frequencies — above about 150 MHz — the coil should be air-cored; below 150 MHz, it needs to be wound on a dust-iron toroid. Some examples of suitable coils for frequency ranges as stated • 150-300 MHz: 50 mm enamelled copper wire, SWG20 (1 mm dia.), one turn; • 75-150 MHz: 9 turns 24 SWG (0.6 mm dia.) enamelled copper wire on a Type T50/12 toroid; • 7.5-15.0 MHz: 70 turns SWG 30 (0.3 mm dia.) on a Type T50/2 toroid. Although the Type BB106 varactor can be used right across the fre- quency range, a Type BB105 is better if most of the work is carried out above 100 MHz, while a Type KVI226 is preferable below 20 MHz. Modulation Frequency-modulating the oscillator signal is achieved by applying the modulating voltage to the wiper of tuning potentiometer Pi via a series resistor and coupling capacitor. It is possible to add a potentiometer for adjusting the level of the modulating voltage, ie. the frequency deviation. Amplitude modulation could be ar- ranged by injecting the modulating signal into gate 2 of the oscillator. This is, however, not a satisfactory method because the internal capacitances of the MOSFET vary with the modulating voltage, resulting in not only amplitude modulation, but also frequency modulation, of the oscillator signal. It is, therefore, better to modulate with the aid of an additional MOSFET connected between the oscillator and the buffer. Fig. 6. Several of these circuits may be connected in series to obtain switch-selected stepped attenu- ations of, say, 2 dB, 4 dB, 8 dB, and so on. Note, however, that the greater the attenuation, the more attention should be paid to screening and decoupling. Any signal "leaks" at the output at low levels spoil the ac- curacy of the attenuator. The table accompanying Fig. 6 gives calculated values for the attenuator resistors; in practice, the nearest standard values in the E12 or E24 series should be used. Note that wirewound resistors should never be used in RF circuits owing to their high self-inductance. }B:BL Output attenuator It is very useful in many applications if the output signal can be attenuated in suitable steps. A suitable circuit for a one-step attenuator is shown in 3-50 selex-io Digi-Course II (Chapter - 4) The internal working of the logic gates inside a Flipflop is quite complex, as we have seen in the last chapter. However, there is nothing to worry about, because once all this complex circuitry is put inside an 1C, we are concerned with only the external connections. These external connections, and the logical behaviour of the Flipflop is all that we need to know, when we are using the Flipflop. We have two sockets provided on our Digilex board for the Flipflop ICs 74LS76. These are marked 1C 6 and IC7. These ICs are quite inexpensive and you can obtain them from any good electronic components hop. Each of these ICs contains two Flipflops and thus we have four universal Flipflops available for experiments. For studying the properties of these Flipflops we can connect the circuit shown in figure 2. A Flipflop made of two NAND gates is used at the input to the clock (CLK) pin of the Flipflop FF1 (half of 1C 6). The NAND Flipflop is used for obtaining noise free clock pulses. These pulses are indicated by the output indicator LED C. Terminals S and R are alternately connected to the ground line to generate the clock pulses. The supply pins of 1C 6 and 1C 7 are not connected to® and(o)on the Digilex board. They need external connection to these supply lines. Output indicator H is connected to the pin Q of the JK Flipflop from 1C 6 to show the state of the JK Flipflop. Pins J and K are used to select the J/K combination at the input of FF1 . Wnen Pin R is momentarily connected to Ground line it gives a 0/1 combination at the input R/S of the NAND Flipflop and sets that Flipflop. This is indicated by the glowing LED indicator C. This high level appears at the clock input of FF1 . Now you can connect the pins J and K to get either a 0/1 or a 1/0 combination. During this, the Flipflop FF1 is unaffected because it has a '1' on its clock input. After setting the J/K combination to 0/1 or I/O, touch the S terminal to the Ground line. This resets the NAND Flipflop and its output becomes 'O', (observe the LED C). This negative going edge at the clock input triggers the Flipflo FF1 and it latches the 0/1 or 1/0 combination which was present on the J/K inputs at that moment. In short, we can describe the above operation as follows : The Flipflop FF1 latches the input combination J/K into the output Q/Q on the negative going edge at the clock We have just seen the effect of setting up J/K either as 0/1 or 1 /0. Now let us find out what happens when J/K is 0/0 or 1/1. For this, first reset the NAND Flipflop. Then set the J/K inputs as 0/0 and clear the Flipflop FF1 by connecting the CLEAR pin to ground momentarily. This gives a 0/1 at Q/CT output. If the NAND Flipflp is now set and reset using the terminals R & S, it will produce a clock pulse at the clock input of FF1 . Note that the Flipflop FF1 remains unaffected and retains its state. Repeat the same experiment with J/K = 1 / 1 . This time, the Flipflop FF1 changes its state on every negative going edge at its clock input. Figure 3 shows the timing diagram of levels at the clock input and the outputs Q and 5 of the Filpflop FF1 (Figure 3) “iXLTLTLTL If you observe the relation between the pulses available J at the outputs Q andU and the input clock pulses, a very | interesting point can be noted. The input pulses are exactly halved in the output, or in other words, we have just covered a circuit which is a 2:1 divider. It is quite obvious that if we feed the output of the first Flipflop to the clock input of another Flipflop and keep its inputs J/K as 1/1 again, we will have a 4:1 divider. Using all the four Flipfolps available to us we can generate a divider chain with 2:1, 4:1, 8:1, and 16:1 divided outputs. This arrangement is shown in figure 4. Note that all the four CLEAR inputs are connected together. This can be used to clear all four Flipflops before we start giving the input >3.5" “ selex As we have a chain of dividers which divide the incoming pulses by two at each stage, the ratios we obtain are all binary values. Though this is quite natural in digital technology, it becomes a bit inconvenient in actual practice when we work with the decimal system. A decimal divider would be of much more value than a binary divider when we are working with the decimal system. This is possible if we take the help of the CLEAR inputs we have connected together. This arrangement has one disadvantage, which can be clearly seen from figure 5. The LED 'E' glows during 9th and 10th pulses and remains off during the first eight pulses. This defect called non symmetrical duty cycle can be rectified by modifying the circuit again as shown in The Digilex-PCB is now available! The Digilex-PCB is made from best quality Glass- Epoxy laminate and the tracks are bright tin plated, the track side is also soldermasked after plating. Block schematic layout of components and terminals is printed on the component side. Price: Rs. 85.00 + Maharashtra Sales Tax. Delivery charges extra: Rs. 6.00 Send full amount by DD/MO/PO. Available from: precious p ELECTRONICS CORPORATION 11. KILN LANE. OFF LAMINGTON RD. BOMBAY - 400 007. selex I 3.53 selex I ^ k i » Filter Capacitor The filter capacitor Cl reducing the voltage variation caused by these half waves. During the very first half wave, the eletrolytic capacitor Cl gets charged to the voltage supplied by the bridge rectifier. When the bridge output starts falling along capacitor supplies some of its stored charge. Thus the voltage at the output of the bridge does not fall as rapidly as it would have done in absense of the filter capacitor Cl . The voltage pattern across the capacitor C-1 is snown in figure 4. The small fluctuation that still exists in the output voltage across Cl is called the ripple voltage. Transistor T1 further reduces this ripple voltage. The part of the circuit that follows Cl is used to obtain variable voltage at the output across capacitor C2. The voltage supplied at the collector of the transistor T1 is always the same as that across the t. capacitor Cl. As we require an adjustable voltage at the emitter, the collector emitter junction must take up the excess voltage. This is achieved by using the property of the base-emitter junction. The.base-emitter voltage remains fixed at 0.6 Volts when the base-emitter junction is forced into conduction. Using this physical property of the voltage at the emitter with respect to ground will depend on the base voltage with respect to ground, (see figure 7.) If voltage, the output voltage at the emitter will automatically change. This means that we must have an adjustable voltage potentiometer PI is used along with two more filter capacitors C3 and C4. Zener diode D1 provides a stable reference vbiiage across the potential devider potentiometer PI (see figure 5 and 6.) A 16 V zener is used in this case, so that a stable 16 V DC is available across the potentiometer PI . The sliding contact of the potentiometer can take voltages from 0 to 1 6V depending on its position. Now once again refering to the figure 7 we can see that the output voltage Ua will be less than the voltage Ub at the sliding contact of the potentiometer by 0.6V. The relation between the two voltages is as follows : Ua = (Ub-0.6) V The output voltage will thus be adjustable by changing the setting of the potentiometer. This relation also explains why we need a zener voltage of 1 6 V to achieve a 0 to 1 5V range at the output. (To be precise, the output will be 0 to 15.4V), When Ub is less than or equal to 0.6V will not conduct and the transistor T1 will be cut off. There will be no output voltage available in Construction Details The circuit described above can be constructed as per the component layout shown in figure 8. Follow the usual sequence for soldering various components. First the jumper wires, then resistors, condensors and semiconductors. Except for resistors, other components in this circuit are polarised. They must be mounted with the correct polarity to avoid any undesired damage. The plus pole of the Zener diode coincides with the ring printed on the body. Since the transistor conducts the entire load current through its collector-emitter it will become hot during operation. A cooling fin or heat sink must be provided for the transistor T1 for proper heat dissipation. selex 1 Table 1 Components R1 22011 MW La : Indicator lamp (230V) SI - Double pole mains switch, 1 Voltmeter (0 to 20V) - Optional 1 Ammeter |0 to 500 mA) - Optional i 3.55 selex selex Figure 2 shows a typical situation where the DC Voltage is being measured across one resistance of a potential divider. As the total voltage across (R 1 + R2) is 9V. we can accurately calculate the voltage available across R2 U2 = U • = 9 V R2 R1 + R2 100 k ' 1 00 k + 10k If we measure this voltage using a multimeter on its 10V range, with an input resistance of 1KS1/V. the reading given by the multimeter will be 4.29 V, instead of 8.18V Surprising! Not so surprising, if we see what effect the input resistance is having on the voltage measurement. The input resistance c; IKfl/V on the 10V range means that we have effectively a 10K!l (Re) resistance in parallel with the 100K11 resistance R2 in figure 2. Thus a total resistance Rg given by R 9 = R2 • R e R2 + R e 100 kfl- 10 kS2 100 kJ2 + lOkfi is introduced in the potential divider circuit. The voltage U2 will now change to 10 kU + 9,09 kJ2 have been 8.18V. From the above calculations it becomes quite clear that the input resistance of the multimeter plays a very important role in deciding the accuracy of reading. Once again refering to the circuit of figure 2. we can observe that if the input resistance of the multimeter was considerably high compared to R2. it would accurate reading. The Circuit Now that we have seen the effect of input resistance of a multimeter on the voltage measurement, let us find out how we can increase the effective input resistance of the multimeter. One such circuit which effectively increases the input resistance of a figure 3. The field effect transistor T1 is the most important component in this circuit. The FET (Field Effect Transistor) used here is N-Channel barrier type. Going into the theory of operation of the FET is beyond the scope of this article. The only important fact to be noted here is that an FET has three terminals called Gate (G). Drain (D) and Source (S) The internal resistance between the Gate and the Source is very high, and its normal value is few Giga ohms (1 Giga ohm = 10 9 il) Thus the circuits using FETs have a very high input resistance. The part of the circuit which decides the effective input resistance of the multimeter attacment is shown seperately in figure 4. The resistance shown as rGS corresponds to the Tnternal Gate to Source resistance of the FET This resistance appears in series with the externally connected resistance R5 multimeter reads as U2. The measurement is totally misleading as the actual value of U2 should combination of R5 and rGS appears is parallel with resistance R3. As rGS is very high compared to R5 and R3. the effective 3.57 selex resistance fo the combination of R3 in parallel with (R4 * rGS) is almost equal to R3. The input resistance of the attachment thus becomes very high and can be calculated as follows : R2 + R3 = 1.1 Mil in the 3V range R1 + R2 ♦ R3 = 4.4 Mil in the 12 V range R3 * R4 = 11 Mil in the 30V range. The FET T1 functions as a Source Follower. The voltage on the source terminal follows the voltage available on the Gate terminal. The amplification factor in this configuration is almost equal to 1. However, as we are not interested in any signal amplification from this circuit, it is of little importance. The FET here has the function of increasing the effective voltage at the input of the circuit to the output without drawing a high load current from the voltage under measurement. The multimeter is connected between the preset PI and potentiometer P2 as shown in figure 3. The components T1, R5, R6, and P2 form a bridge circuit. (Refer to page 1.65 of our January 1986 issue). Preset pot PI and the multimeter form the middle branch. By adjusting the setting of* potentiometer P2, the multimeter reading can be set to zero volts, when no input is present. The function of PI is described The diode 01 protects the FET against negative voltages on its Gate. This prevents any damage to the FET in case the voltage under measurement is connected to the input with a reverse polarity. The diode never allows the Gate voltage to fall below 0.7V. Construction Details A SELEX PCB will have enough space to mount all components of the attachment, including a 9V battery pack. The component layout is shown in figure 5 All the usual precautions and rules of construction should be properly followed. Special attention must be given to the terminals of FET and the polarity of the diode, the battery pack can be placed on the free area of the PCB. A rubber band can bu used as the battery clamp as shown in figure 6. This is possible by using two bent soldering pins on each side of the battery pack and then attaching the rubber band through them. This will securely hold the battery pack in its place. Compensation construction is complete, it can be tested for proper operation. The voltage on the Source terminal of the FET should be approximately 2V with no input voltage connected at the input terminals of the attachment. If this voltage is correct, a multimeter with an input resistance preferably around 20Kf!/V can be connected in its palce with correct polarity. With no input voltage present at the test terminals of the attachment, the multimeter reading can be set exactly to zero volts by adjusting the potentiometer P2. It is not enough just to set the zero reading The full scale reading also must be correctly compensated. For ad|usting the full scale deflection, we need accurate voltage reference of 3V. 12V or 30V The reference voltage can either be obtained from a good variable voltage supply, or 3V can be ontained using 1.5V battery cells. (These cells should be brand new. so that they really give 1 5V If a good variable voltage supply is not available for the compensation adjustment, one has to accept the slight inaccuracy that may result from using battery cells as reference voltage. The reference voltage available is connected to the corresponding input socket. The multimeter is set on the IV or 2V range and the preset pot PI is now so adjusted as to get the needle of the scale deflection mark, of that range. Once this compensation is done, it also holds good for the other two ranges. When using the multimeter with the attachment in future, it must be set to the "range (IV or 2V) on which it was compensated. Comparison The attachment circuit described here was actually tested and the reading accuracy was compared using a highly accurate Digital Voltmeter (DVM). the deviation in readings were observed to be as follows. 0.19% deviation on 3V 0.26% deviation on 12V range 1.11% deviation on 30V range. The average deviation thus lies at about 0.52%. Quite an accpetable tolerance, considering the low cost of the attachment circuit. 3-58 ORION located i computer :tly processed information Rajkumar Engmee 106. Ba/sons Ind. Chakala Road. Andheri I East I Bombay 400 099. DIGITAL WATTMETER Economy's Digital Wall DELEGATED INDIAN ENGINEERING COMPANY PLASTIC MOULDED DOUBLE DIN STD INSTRUMENT CASE "COMTECH" — T-192 is a double DIN standard Instrumer Chips makes POWER PUSH BUTTON SWITCH Rajkumar engineers offe For futher details, please contact : HINDUSTAN INSTRUMENTS LIMITED 603 Vishal Bhawan 95 Nehru Place Bombay The switc amperes PROXIMITY SWITCHES V Aplab is the largest private manufacturer and exporter of professional electronics equipment small India. Aplab specializes quantity non-consumer products and a wide variety of custom made equipment. Aplabs' wide range of standard products include ajP based instrumentation systems. GPIB based instruments. Oscilloscopes. Signal Generators. DC Power Supplies. setters. Aplab is serving the nation by catering the vital needs of Defence. P &T, Railways, Research Centers and all the major, public and private sectors. Aplab has a strong R & D base with more than 200 qualified Engineers and Managers and welcomes R & D projects. Applied Electronics lid. A-5/6. Waqle Industrial Estate. Thane 400 604. Bombay. Indi 161/62/63 Tlx: 01 1-71979 APEL IN. Cable: APLABTHANA ■ if r * — * IC SOCKETS HIGHLY RELIABLE CONSISTENT SUPPLY AT AFFORDABLE PRICES t £hampmn ^lectpanies Put Ltd Champion Electronics Pvt Ltd. S-1 7. MIDC. Bhosari. Pune 41 1 026. INDIA Phone (0212) 83791. Cable : CHAMPION Telex 0145-333 MCCI IN. 0145-505 MCCI Socket all your IC’s with CHAMPION IC Sockets and MULTIPLY* your PROFITS 1 1 Plots of Spring Contacts ore Oriented to Flats of IC Leads. Duol Leaf Spring Contacts. profile -3,94 1 roll height. For details write to us Advertisers Index classified ads Available rechargeable torch using beam uplo 200 feet. Operates two hours, when fully charged. Send Rs. 225/- by M.O./D.O. Postage Free. SUN ENTERPRISES, 56/1 , Industrial Estate. Vithalwadi. Bhavnagar 364 001 . Walkie talkie transmitter circuits avai- lable. Sample circuits Rs 1 5/- Write to RND Electronics, 12-11-1313. Ragha- vendranagar, Secunderabad - 500361 8085 Microprocessor trainer in brief- case Rs. 2970/- Built in EPROM PROGRAMMER power supply 2K CMOS RAM with dry-cell back up expandable to 8K. 12K, user EPROM New Age Electronics third floor, Laxmi Mahal, Near Vandana Cinema. Agra Road. Thane-400 602. READ all ELECTRONICS BOOKS at 1 /3 rd of original price. For details contact: SRI KRISHNA ASSOCIATES. B/25. Mahesh Co-operative Housing Society Dattapada Road Bombay - 400 066 ADVANI-OERLIKON 3-11 ACE COMPONENTS 3-10 APEX 3-71 APLAB 3-63 ATRON ELECTRONIC 3-68 BAKUBHAI AMBALAL 3-14 . CHAMPION ELECTRONICS . 3 69 COMPONEN T TECHNIQUE .. 3-12 DEVICE ELECTRONICS 3-13 DYNALOG MICRO SYSTEM . . 3-76 ECONOMY ENGINEERING . . .3-12 ELCOT 3-07 GALAXY ENTERPRISES 3-08 IEAP . 3 06 INDIAN ENGINEERING 3-10 KLAS ENGINEERING 3-08 KOENIG ELECTRONICS 3-10 LEADER ELECTRONICS 3-08 LEONICS 3-70 MOTWANE 3-67 OSWAL ELECTRONIC CO. 3-02 PADMA 3-66 I PIONEER 3-06 PRECIOUS KITS 3-62 J RAJASTHAN ELECTRONICS CO. 3-02 ROCHER ELECTRONICS 3-06 SAINI ELECTRONICS 3-70 SIEMENS 3-61 SUCHA ASSOCIATES 3-68 TESTICA 3-71 TEXONIC 3-70 UNLIMITED ELECTRONICS . . . 3-68 VASAVI 3-12 VISHA 3-75 YABASU 3-04 ZODIAC 3-72 □□□□□□□□□□□ □□□□□□□□□□□□ □ □□□□□□□□□□□■ □ □□□□□□□□□□ 1 □□□□□□□□□□a 3-74, ft.N No. 3 9 881 7 83 MH/BVW-228 L1C. No. 9 1 Dynalog tUcro-Systems institutions. Under ssor and Computer the Microprocessor iner Kits based on s interfacing and per‘P' syllabus prescribeo et your need. , ,he Microprocessor achment Organisation Laborat ailed 'Super SO' with IBM PC/XT. Using' inisation Laboratory- sslully conduct these systems prescribed syllabus. to your reply- Thanking you anc Yours falthlully. For Dynalog Mic FOR ENGINEERING COLLEGES AND TECHNICAL INSTITUTIONS ’