automatic | squelcu « erasing EPROMS selektor Express train to nowhere. 3-13 high com monitor extension 3-14 Very few additional components are required to extend the High Com and provide multiple head tape systems with a monitor facility. wind sound generator 3-16 Anyone requiring a windlike sound, such as amateur photographers during a film or slide show, can make good use of this electronic wind sound generator. adding the finishing touches to the NEW Elektor synthesiser ... 3-18 The final article on the basic version of the NEW Elektor synthesiser describes the control and output module (COM) and a simple power supply. automatic squelch 3-26 Besides the straightforward construction and calibration involved here, the automatic squelch has a major advantage: You do not need to be an expert to install it into the audio section of a receiver. DNR printed circuit board 3-28 Last month we promised a practical noise reduction system - and here it is. The circuit literally makes noise go 'off the air'. lead acid battery charger 3-32 The circuit presented in this article not only charges lead acid batteries, but also acts as a power supply. It is polarity-protected and includes current and voltage limiting. It also provides charge control and a polarity indicator. In other words, the battery charger is practically foolproof! blue LEDs 3-36 During the past three years, the physical properties of SiC (silicon carbide) have been the subject of an enlightening study. This article describes this semiconductor material and shows how blue-emitting SiC diodes can be manufactured. missing link 3-41 AD/DA conversion 3-42 T. Schaerer Analogue to digital conversion technology has advanced to the degree at which electronics enthusiasts can afford to experiment with digital cir- cuits, such as the one described in this article. dissipation limiter 3-47 The circuit illustrated here provides a safeguard for the 2N3055, a sort of life insurance. EPROM eraser 3-50 Since Elektor has paid a good deal of attention to EPROM programmers lately, it is high time a suitable EPROM eraser was considered. The ultra- violet method described here is both efficient and fairly cheap. polyphonic synthesiser 3-51 The advantages of computer controlled polyphonic keyboards are dis- cussed in this article, which at the same time paves the way for the printed circuit boards and the constructional details to be published in due course. market 3.54 elektor march 1982-3-13 Express train to nowhere stores electrical energy Experts throughout the world agree that the future for energy would look much smoother - and renewable energy sources would look more tempting - if the basic problems of bulk energy storage could be solved. Even in existing centralised energy utilities, and particu- larly electricity grids, the ability to store energy at times of low demand for use during periods of peak consumption would have immediate and substantial benefits. Generating equipment of whatever kind could be operated at its optimum load at all times and the ironing out of the peaks and troughs from the demand curve would mean very large reductions in capital and operating expenditure. In most European countries and the United States of America the possible savings approach 40%, although 25% might prove more realistic. Troughs in demand An advance of this kind would require storage systems that operate on the basis of electrical input and electrical output, and with a capacity commensur- ate with that of utility power station production — perhaps in the range of 500 ... 1 000 megawatt hours. The sys- tem would also have to operate at high efficiency over a storage period of 12 to 36 hours, to take into account daily and weekend demand troughs, and be capable of being constructed close to the power station for which it was to serve as the store. In areas where suitably large lakes can be created by building dams, water can be used in pumped storage systems with a storage time-base that is virtually unlimited. However, overall efficiency is not particularly high because of losses in the pumping process, and in any case pumped storage, like the storage of energy in the form of high pressure air underground caverns, is feasible in only a few localities. During the past couple of decades a number of engineers and scientists throughout the world have attempted to find a solution to the problem, and many proposals have involved the use of flywheels of one kind or another. This is because, in broad principle, the flywheel is an excellent storage system, capable of being driven up to great speeds by high efficiency motors which can also serve as very efficient generators to take the stored kinetic energy out again. But conventional flywheels on a central spindle have basic design problems which become increasingly prohibitive as the scale — and hence the potential storage capacity — is increased. The ideal flywheel would have all its mass at the outer rim, which is the point of highest velocity. But the larger the mass at the rim and the higher the velocity, then the higher the structural stresses between the spindle and the rim and the greater the proportion of mass of structural material that has to be introduced into the low velocity area. Exotic and expansive Even with the most exotic and expens- ive materials, a store of 200 megawatt hours is approaching the practical limits. Smaller stores exist and have operated well for many years — Professor Oliphant's homopolar motor-generator in Australia is a famous example — but these are not at utility scale. However, two British scientists. Dr F. M. Russell of the Rutherford and Appleton Labora- tories and Dr S. H. Chew, who worked until recently at the University of Malaya but is now at Oxford University, have come up with a genuinely revol- utionary idea for, in effect, they have turned the flywheel inside out. According to Dr. Russell, when the problem is thought through, it becomes obvious that you have to do away with the central spindle and all the engineer- ing difficulties it entrains. So they de- cided to design a flywheel with all the mass in the rim and the bearings outside. The result looks highly promising. Their basic idea, which already has the interest of engineering and construction companies in Britain and is being evaluated in considerable design detail, is deceptively simple. In effect it in- volves the construction of a high speed underground railway with a single 'train' occupying the full length of a circular track. Taking a diameter of 1000 metres as a design criterion — for this fits comfortably within or just around large power station sites — the scientists have examined the problems and hence the required technology for a 500 megawatt hour store. The technology is eveilable This was described in detail at the recent Second International Conference on Energy Storage at Brighton, England, and the specification contains several surprises. The first, and perhaps the most important, is that almost all the required technology is already available because it has been developed in several national programmes for high speed trains. The tunnelling technology, devel- oped both for urban subways and other transport requirements, and in a high precision form for large 'atom smashing' accelerators — such as the 27 kilometre diameter electron-positron collision ring soon to be built in Geneva — is also at the required level. Even more encouraging are the first cautious, broad assessments of cost. They suggest that the underground train storage system should not be very different from that of pumped storage systems. Fears that rolling friction losses and losses caused by air resistance in the tunnel would prove serious were shown to be unfounded as the study pro- gressed. The system emerged as a train, carrying a mass of dense material - the heavy rock excavated during tunnelling could provide the bulk of the mass - driven by 24 motor-generators and borne by tracks taking both vertical and horizon- tal loads. As the detailed engineering evaluation was pushed forward the basic principles were clarified but remained unchanged. The required storage capacity could be attained at train vel- ocities only about twice those becoming common in high speed rail systems. The 'reference' design has a maximum linear speed of 300 metres/second, a design efficiency substantially greater than that of pumped storage over a period of 24 hours, and the capability of being built anywhere since the sub- surface geology is sufficiently strong to accept the transfer of large inertial, cen- trifugal and gravitational forces. The 'switching time' of the system - that is from store to energy output of 200 megawatts, is assessed at 3 milli- seconds, which is adequate for the most sensitive grid control systems now envisaged. Wheel and track wear But detailed study of the reference design has also revealed problems, and the most serious involves 'bearing' — that is wheel and track - wear. One of the design requirements is that the storage system should have a virtually maintenance free life comparable to or greater than that of the power station with which it would be associated. That means 30 years or more, and to achieve that lifespan the loads transmitted to the tracks would have to be reduced by 'an order of magnitude or a little more', according to Dr Russell. The inventors claim that the technology already exists for the provision of this kind of magnetic load reduction but prefer not to talk about it yet. The system they have in mind would be 'advanced' in the sense of concept, yet very robust. Costs are uncertain at this stage, but there is no reason to believe that they would be outrageous. Industry is interested and it seems quite feasible that the expense will still be within striking distance of the costs of pumped storage. Anthony Tucker, Science Editor for 'The Guardian', London. (744 S) 3-14 -ele 1982 high com high com monitor extension . . . for multiple head tape recorders We have been asked by a number of readers whether it is possible to extend the High Com circuit so that it can be used in conjunction with the monitor facility found on multiple head tape recorders. Initially, this came as rather a surprise, since the High Com system was designed for 'normal' cassette decks. Nevertheless, it is possible to extend the system so that the monitor facility can be fully utilised. Many readers may not have yet had the opportunity to construct the noise reduction circuit published in the March 1981 issue of Elektor, if not . . . now's your chance! Others may wish to extend the 'old' circuit. In either instance the Elektor High Com system will have to be available in its original form to start with. However, before we continue, let's make a study of some of the background details so that we 'know what we are doing'. Tape recorder technology Reel-to-reel tape recorders and cassette decks can be placed into two main categories: those with monitor facility and those without. In principle, three heads are required: an erase head to 'wipe' the tape clean; a record head to transfer the relevant signal to the tape; and a playback head to retranslate the recorded information into an electrical For reasons of economy, the record and playback heads are very often combined into a single unit. It should be noted, however, that a record/playback head cannot record any kind of signal and play it back (monitor) at the same time. Recordings can only be monitored if separate record and playback heads are available. Supposing, for instance, that a noise reduction system was connected be- tween the signal source and the record head, and the playback head was used to monitor the recording. Then the same noise reduction system would have to be connected between the playback amplifier and final output medium. Since the High Com system can only be operated in the 'record' or 'playback' mode, separate record and playback channels will have to be added, in other words, two noise reduction systems! The extra expense can, of course, be avoided by simply playing back the companded signal, but this will not guarantee high quality reproduction. The monitor Fortunately, very few additional com- ponents are required to extend the High Com system. Firstly, two more modules will have to be added: one for the right- hand channel and one for the left-hand channel. Since the recording channel constitutes the most complicated circuit, it is included on the existing board. The majority of the playback channel, on the other hand, consists basically of the High Com module. To work out the best method of constructing the monitor section, let's take another look at the circuit diagrams in figures 6 and 7 of the original article published in the March 1981 issue of Elektor. One solution, for readers with plenty of time and money, is to build the complete device twice and record through one and monitor with the other. However, there are cheaper and less time-consum- ing methods, which will be discussed here. The circuit diagram of the prototype is shown in figure 1. The playback channel consists of the High Com module, the input and output interfaces and the electronic switches. For the monitor channel, the interfacesand the electronic switches can be omitted if desired, in which case the tape unit will be in the High Com mode permanently. This option is recommended as it enables any differences in level to be equalised from the start. Construction For those readers who have not yet built the original Elektor High Com system, full constructional details will be found in the March 1981 issue of Elektor. As far as the monitor extension is concerned, two extra High Com modules are required together with the components listed in table 1 . These components are the same as those used in the original playback system and Table 1. Parts list for figure 1 Resistors: R19.R119 = 82 k R20.R120.R23.R123 - 47 k R21.R121 - 10k R22.R122- 15 k R24.R124.R25.R125- 5k6 R54.R1 54 - 100 k P1.P101.P2.P102 = 25 k preset Semiconductors: IC3 - MC 14066. CO 4066 IC4- RC4558P all other components are included on the High Com modules. should be mounted on a suitably sized piece of Veroboard according to the circuit diagram in figure 1. Solder pins should be provided for each of the connection points, the ones used to mount the High Com modules should be 1.3 mm in diameter. Of course, an extra main board could also be used and the superfluous components omitted, but this may prove to be rather expens- ive. The Veroboard should be the same width as the main board. This ensures that there is plenty of room for the two modules, which can be mounted at right angles to those on the main board. This allows the various connections to the main board to be situated along one of the sides of the extension board, while all the external connections can be situated along the opposite side. The extension board should be pos- itioned so that the two sets of con- nections marked; 'S4a', '+15 V', 8 V', '+8 V', 'ground', 'S4c', 'S2' and 'P' are exactly opposite the outputs on the main board. As a result, the intercon- nections can be kept as short as possible. As far as calibrating the circuit is concerned, the same procedure as that described in the March 1981 issue of Elektor should be followed. References: Noise reduction, Elektor February 1981, page 2-04. The High Com noise reduction system, Elektor March 1981, page 3-06. M 3-16 — elektor ch 1982 a simple effects unit Generating wind at professional film and television studios is a relatively simple matter: all they have to do is press a button and a powerful fan supplies anything in the way of simulated sea breezes to gale force winds. In the home, such effects are much harder to create, and usually result in the perpetrator being thoroughly winded . . . Anyone requiring a windlike sound, such as amateur photogra- phers during a film or slide show, can now make use of this portable electronic wind sound generator. A few components, a battery and an amplifier are all that are required to produce effects ranging from a gentle breeze to a Caribbean hurricane. Just the thing for livening up a dull party! H. Pietzko The sound of the wind is very similar to the major headache of HiFi enthusiasts, noise. Nevertheless, it is not sufficient to utilise just a noise generator to imitate gusts and gales, especially as the main characteristic of the latter is to produce considerable volume within a limited frequency range, although the complete audio spectrum is represented in the signal. The increase in volume accompanied by a howling or whistling tone is caused by diverting, compressing and then expanding the actual wind. The slightest alteration will produce a different sound. Of course, the same principle applies to wind instruments where the 'column' of air inside a 'tube' is compressed and expanded to obtain the various notes of the scale. Electronic wind We are not going to discuss electronic wind instruments here, as the majority of music synthesisers are able to imitate them. Rather, we are going to discuss an effective wind sound generator which uses a reverse biased germanium diode as a 'noise generator’. The block dia- gram of the unit is given in figure 1. If only a small current is allowed to pass through the diode the current will not remain stable. At room temperature (about 300° Kelvin), which is a very high temperature for diodes, the elec- trons in a crystalline structure move about in a totally random manner. They do not become immobile until the temperature drops to 0° Kelvin, the absolute zero level. This (normally undesirable) effect, which manifests itself in the form of audible broadband noise, is eminently suitable for this particular application. After being ampli- fied by a large amount, the noise signal can be further 'processed'. The most straightforward method is to use a bandpass filter which greatly amplifies part of the noise spectrum, as shown in figure 1. The bandwidth of the filter must be very narrow in order to achieve maximum performance. In the design presented here, the selectivity (Q) and the centre frequency of the filter are variable, enabling a large variety of 'wind sounds' to be selected. There is no need to worry about win- ding an inductor, for the parallel tuned circuit indicated by the bandpass filter section of figure 1 , as the filter is con- structed around two opamps. The circuit The circuit diagram of the wind sound generator is shown in figure 2. The germanium diode D1 and resistor R1 constitute the actual noise generator. The noise signal is amplified by opamp A1 to produce a noise level of about 150mVp p at the output (pin 1). The amplified noise signal is then fed through a high pass filter consisting of resistor R4 and capacitor C4 and then through a low pass filter comprising R6/C5 and R7/C6 to reduce the bandwidth. The circuit around opamps A2 and A3 forms the 'variable inductance' for the bandpass filter. Inductors can be 'imitated' by using a capacitor and a gyrator, as has often been done in Elektor circuits in the past. A different approach involves two opamps. Resistor R8, capacitor C8 and the 'coil' (A2/A3) form a tuned circuit with a resonant frequency that can be adjusted by means of potentiometer PI. The im- pedance between the non-inverting input of A2 and ground is: Z = jw (PI + R9) • T Thus, the inductance will be: L = (PI + R9) • T where T = RIO • C9 = (P2 + P3) • CIO The inductance of the 'coil' and there- fore the centre frequency of the bandpass filter can be adjusted by means of potentiometer PI. The Q of the filter can be regulated by means of P2 and P3. As a result, the wind force is established by the former and the volume of its whistling tone is established by the latter. Opamp A2 also acts as a buffer stage and provides a low impedance output for the wind signal. The amplitude at this output will only be about 1.4 mV, o-DHg-H- 14 ~j> > ~ o1 wind sound generator Figure 1. The block diagram of the wind sound generator. therefore the signal needs to be ampli- fied somewhat. This is accomplished by means of opamp A4, the final amplitude of the wind signal being in the order of 100 mV. Construction, calibration and operation Although the circuit has very few I components, the performance is quite surprising. All the components (apart from the potentiometers) can be mounted on the printed circuit board shown in figure 3. Since the current consumption of the circuit is a mere 8 mA, it can be battery powered. A separate small power supply could also be used provided the supply voltage is adequately smoothed. A number of suitable circuits have been published in Elektor over the years. Calibration simply involves the adjust- ment of preset potentiometer P3. With PI and P2 set to their minimum and maximum resistances, respectively, P3 is turned (starting from its minimum resistance value) until the bandpass filter is just about to change frequency. In other words, the amplifier and loudspeaker should not emit the slightest breeze I It may be advisable to connect the wind sound generator to a mixer prior to the audio amplifier. This would enable the unit to be operated with maximum efficiency during slide and/or film shows etc. The device is, of course, also suitable as a sound effects generator, in which case it can be connected directly to the line input of the audio amplifier* 3-18 — ' 1982 NEW Elektor synthesis The printed circuit board for the Formant COM module, published in April 1978, can be used here with no modifications, although not all the cop- per tracks need be used. The circuit in- cludes bass, middle and treble controls, a sub-sonic high pass filter, a preset gain facility and a master volume control. The complete circuit diagram of the COM module is shown in figure 1 and the wiring connections for the printed circuit board are given in figure 2. Only four pins of the connector are actually required in this instance. These are: adding die finishing touches to die IVEW Elektor synthesiser the COM module, the power supply and a few constructional hints The final article on the basic version of the NEW Elektor synthesiser describes the control and output module (COM). This was originally designed for the Formant synthesiser and was fully described in the April 1978 edition of Elektor (page 4-33). It includes a preamplifier with bass, middle, treble and volume controls. The power supply for the synthesiser is very simple and consists of virtually only two voltage regulator ICs. ground; the positive 15 V supply rail; the negative 15 V supply rail; and a signal input, which is connected to the output of a VCA. The tandem poten- tiometer Pla/Plb prevents the remain- der of the circuit from being over- modulated and at the same time ensures that the desired signal is not 'drowned' by noise from the circuitry shown between Pla and Plb. Depending on the settings of the various synthesiser controls, a brief low fre- quency signal produced when a key is depressed could cause damage to the loudspeakers. Such detrimental tones are suppressed by means of the low pass filter connected in front of the tone control network. The filter has a cut-off frequency of about 20 Hz and is similar to the rumble filters found in stereo equipment. The level of treble and bass is adjusted by means of a 'Baxandall' network con- structed around opamp A2. The output of the Baxandall stage is fed via a buffer amplifier to a separate ‘pre-emphasis' circuit constructed around opamp A3. This section of the circuit controls the 'middle' frequencies. The gain of the output stage, A4, can be adjusted by means of preset poten- tiometer P5 between a factor of 1 .8 and 1 1 times depending on the input sensi- tivity of the power amplifier connected to the COM module. The output signal from A4 is fed to a jack (or DIN-)socket situated on the front panel of the module. For completeness' sake, the 'old' p.c. board is repeated at the end of this article (figure 10). How to incorporate the COM module The bus boards mentioned in the pre- vious articles on the NEW Elektor syn- thesiser have to be slightly modified in order to accommodate the COM mod- ule. As can be seen from figure 3, the pins of the 21 -way connector soldered to the COM printed circuit board will not fit into the holes of the corre- sponding socket, if the latter is mounted on a bus board that has been inserted in the slide-in unit using the 'standard' method. The pins are positioned exactly half-way between the holes. The sol- ution is to turn the bus board 180° before insertion and to remove the first and last pins of the connector with a pair of suitable cutting pliers. The power supply The NEW Elektor synthesiser requires a power supply capable of producing + and —15 V and which will maintain a load of 200 mA per rail. Furthermore, the polyphonic extension to be de- scribed later requires a +5 V supply. A suitable circuit is given in figure 5 (and a p.c. board layout in figure 1 1 !). Obviously, the components for the +5 V supply need not be mounted yet (IC3 with its heatsink, C7 and C8). Although it is not strictly necessary, it is a wise precaution to mount the volt- age regulators (IC1, IC2 and IC3) on small heatsinks. After all, it is better to be safe than sorry! How to connect the power supply For safety reasons, it is not rec- ommended to mount the power supply transformer directly on the printed cir- cuit board. Having a copper track bear the brunt of 240 volts is rather risky to say the least. The transformer should be mounted on a piece of aluminium, about the size of a eurocard, which will also act as a 'screen' from the rest of the circuit — provided the aluminium is grounded. The power supply and transformer can be wired directly to the connector. A robust, mechanical connection can best be made using long screws and spacers, as indicated in figure 6. 1 Figure 1. The circuit diegrem of the control end output module (COM) is identicel to that used in the Forment design. Two LEDs on the front panel (con- via three supply voltage rails. The signal quency modulate the VCOs, the VCF nected to the + and -15 V supplies) paths are indicated as thick, black or all the modules simultaneously. The allow the user to ascertain at a glance lines. ADSR outputs are linked to the control whether the power supply unit is The output signals from the two VCOs inputs of the VCF and VCA. The KOV working correctly. and the LFO are first fed to the mixer inputs of the two VCOs are linked to input of the VCF, then to the VCA and each other and also to the KOV output finally to the COM unit. The gate pulse of the Formant keyboard (see the from the Formant keyboard also con- article on the VCO published in the Constructional hints trols the vibrato section of the LFO/ December 1981 issue of Elektor, page Figure 7 shows all the basic connections NOISE module, but not the two envel- 12-39). for the various synthesiser modules. The ope generators. The various modules can all be accom- boards are linked to the power supply The LFO signal can be used to fre- modated in a 'card frame'. Suitable / / Figure 2. The wiring details of the COM unit. Elektor synthe 180° degrees before being installed. systems can be obtained from most components retailers. For the sake of clarity, the connections between the printed circuit boards and the front panels have been omitted from the drawing in figure 7, only the links be- tween the individual boards are shown. Figure 8 shows the rear view of a slide- in case with its seven bus boards. Pro- vided the boards are wired from right to left, and each module is checked separately, very little can go wrong. The connecting leads do not have to be in- sulated. The socket for the keyboard connection can be mounted on a small piece of aluminium the size of a bus board. This can be inserted between the power supply and the bus board of the first VCO. A suggested layout for the front panels is shown in figure 9 and it also gives an idea of the required measurements. When inserting the modules into a standard case, make sure that the total front panel width corresponds to the sum of the values indicated on the drawing. To be certain that all the potentiometers fit on the various front panels, miniature types with a spindle diameter of 4 mm should be used. Of course, many readers will wish to design their own cases and front panels, in Figure 5. The circuit diagram of a suitable power supply for the Elektor synthesise adding the finishing touches to the NEW Elektor synthesiser elektor march 1982 - 3-21 which case we would be interested to g hear about the results. As far as legends on the front panels are concerned, the (pre-drilled) front panels can be marked with rub-on lettering (available from stationers and elec- tronics retailers). The panels can then be covered with a thin layer of trans- parent adhesive foil and the various holes cut out with a sharp knife. The foil should be slightly larger than the front panel in question, so that it can be wrapped around it and will not peel off easily. Alternatively, the panels can be sprayed with a suitable laquer after the legends have been applied. With a little time and patience, the panels can be made to look very professional. Principal settings for the synthesiser Now that the NEW Elektor synthesiser has been completed, it is time to try out a few sounds. Admittedly, the choice of modules is rather limited compared to the Formant, but then the whole point of the new system was to make it easier to produce synthesiser music on stage, which meant reducing the vast array of Figure 6. For safety reasons, the transformer is best mounted on a separate piece of aluminium. knobs and buttons used in the Formant 3-22 - elektor 1982 adding the finishing touche i NEW Ele synthesis system to an absolute minimum. The re- maining 28 controls still offer plenty of g musical possibilities. The following settings can be combined as desired: 1 . with or without glissando 2. one or two VCOs 3. in the case of two VCOs: a. both with the same frequency b. with an octave between them c. with a fifth, a fourth or a third be- tween them 4. filter with envelope control a. percussive sounds: attack/decay curves, attack time = 0 b. wah-wah and brass instruments: attack time not equal to 0, ADSR curve 5. filter without envelope control 6. tracking filter 7. VCA envelope: this must be tuned to the VCF envelope. A short VCF attack and decay time will not go into effect, for instance, if the VCA attack time is long. The VCA plays an im- portant role, whenever the filter is not modulated by way of the envelope generator and the cut-off frequency is somewhere in the audio range (see point 5). 8. additional mixing of LFO and noise A few examples: (The names given below to the various sound effects are purely fictional and do not claim to be official terms.) 1. Spherical sound: two sawtooth sig- nals of the same frequency/glissando. Filter envelope set on zero/Q value on Adjust the filter cut-off frequency to allow the entire frequency spectrum to pass/ VCA: attack: zero sustain: maximum release: 1.2 seconds 2. By using two symmetrical VCO squarewave signals while keeping the other modules in the same setting, an effect similar to that in 'Lucky Man' by Emerson, Lake and Palmer is created. 3. Disco sound: VCO setting as in 1/no glissando. Set the filter cut-off fre- quency to zero and the envelope am- plitude to maximum. Adjust the Q Filter envelope: attack = 0, sustain = 0. Using different decay times, a great variety of percussive effects can be produced, some of which sound like the staccato accompaniment often used in disco numbers. The effect is en- hanced by separating the two VCO frequencies by a fifth. Remember that melodies with parallel intervals do not always combine well with accompani- ment chords played on a different instrument. 4. 'Sound the trumpet': VCOs: sawtooth or squarewave, same frequency or a third, fifth or octave interval between them. Filter settings as in point 3. Filter envelope: attack time not equal to zero, sustain equal to 100%, release very brief, but not zero. 5. Woodwind instruments: Figure 8. Res adding the finishing touches to the NEW Elektor synthesiser elektor march 1982 - 3-23 A single VCO with a squarewave Filter envelope: see point 4. Filter envelope amplitude: low. Try out different cut-off frequencies! 6. Sinewave sound : VCO with triangle signal. Switch on tracking filter operation and set the cut-off frequency to match the VCO frequency. Filter envelope = 0 VCA: see point 1 . We will not go into all the possible sound effects that the synthesiser is capable of producing, as this would fill several issues! In any case, it is much more fun to experiment and find out for oneself. After a certain amount of practice readers should be able to dis- cover all sorts of novel and interesting combinations and settings. This ob- viously involves a little more than aimlessly twiddling the knobs. The tones obtained using this method are likely to be cacophonic, if anything. Thus, a systematic approach and fine tuning are an absolute must when oper- ating the synthesiser. This completes the series on the basic version of the NEW Elektor synthesiser. The forthcoming sequel will describe how to construct a polyphonic key- board and how to connect it to the existing modules. Figure 9. A suggested fr panel leyout for the varic 3-26 - ele arch 1982 automatic squelch The audio bandwidth in communi- cations equipment is almost always relatively narrow, which is quite suf- ficient as only information has to be transmitted. This transfer of infor- mation is normally accomplished by means of the human voice. Conse- quently, the chosen bandwidth is sufficient to produce a clearly audible sound and nothing more. Depending on the quality required, the bandwidth is usually in the order of 1.5 .. . 4.5 kHz, which is a familiar value for radio amateurs and CB operators. It is normal for the transmitter to be switched off, immediately after the information transfer has been com- pleted. The noise which builds up during the breaks can be suppressed with the aid of a squelch circuit. Basically, there are three different types of squelch systems: carrier squelch; noise squelch; and signal-to-noise squelch. The carrier squelch circuit derives its information from the pres- ence or absence of the transmitted carrier wave. It is evident that this system cannot be used with single side- band (SSB) or double sideband (DSB) transmissions as the carrier wave is suppressed. The noise squelch circuit checks whether or not the transmitter is active by examining the amount of noise present outside the audio pass band, since a strong noise signal is produced when no transmitter signal is present. The last system is the signal/ noise squelch circuit which determines the relationship of the detected signal to the amount of noise present continu- ously. The audio signal is not passed on to the amplifier stages if the ratio of signal/noise drops below a certain level. The major drawback of this system is that it is a rather extensive and com- plicated circuit, compared to the other systems. At the beginning of this article we mentioned the bandwidth of communi- cations equipment. This will be our starting point, since we are going to describe a fully automatic noise squelch This circuit is primarily intended for narrow band FM receivers (such as CB equipment). The intention is to con- struct a circuit which examines the level of noise present in the audio stages within a small frequency band and just outside the audio spectrum. The signal path between the demodulator output and the audio input is interrupted as soon as the noise exceeds a pre-deter- mined level. Consequently, the loud- speaker will fall silent until an actual transmission is received. The block diagram of the automatic squelch control circuit is illustrated in figure 1. The output signal from the demodulator is fed to a buffer amplifier, A1 . The output of this buffer is then fed back to the volume control (the audio input) via an electronic switch (ESI). However, the buffer output is also fed through a bandpass filter (A2) to an amplifier (A3) and a rectifier stage (A4). The DC output of the rectifier stage determines whether or not the electronic switch, ES4, is open or closed. The latter in turn controls electronic switches ESI and ES2. When the noise level is below the pre- determined value switch ESI is closed and switch ES2 is open. Therefore, the output signal from the demodulator is passed directly to the audio input. On the other hand, when the noise level is excessive, switch ESI will be open and ES2 will be closed. This effectively interrupts the signal path and short- circuits the input to the audio stages. The combination of ES1/ES2 is in- cluded in order to eliminate any dis- turbing switching sounds from the output amplifier. The circuit The circuit diagram of the automatic squelch control is shown in figure 2. The connection to the 'hot' end of the volume potentiometer is broken inside the receiver. This lead is then connected to the input of the buffer amplifier A1. The output of the buffer amplifier is then connected to the 'hot' end of the volume potentiometer via ESI . As the circuit is powered by a single supply rail, the opamps have to be biased 'artificially'. This is accomplished by the potential divider R3/R4, resistor R1 and preset potentiometer P 2. Conse- quently, the non-inverting inputs of A1 and A2 receive approximately half the supply voltage. The output of A1 is also fed to the input of opamp A2, which forms the bandpass filter, via capacitor C4 and preset potentiometer P2. The LC tuned circuit connected between the inverting input and the output of A2 determines the centre frequency of the bandpass filter. The centre frequency can be changed quite easily, by altering the value of the inductor, LI, and/or the capacitor, C5. With the values indicated, the centre frequency is around 5 kHz. The signal level fed to the input of the bandpass filter can be set by means of P2. On its route to the rectifier stage con- structed around A4, the output signal from the bandpass filter is amplified considerably by opamp A3. The gain Figure 1. The block diagram of the automatic squelch control. automatic squelch A squelch ensures that a receiver amplifier does not get inundated by unwanted noise when the transmitter signal is not present. Such a device is essential for communications equipment, since the transmitter is switched off between transmissions. If the receiver does not possess a squelch circuit, the noise literally bursts out of the loudspeaker during these breaks. Besides the straightforward construction and calibration, a major advantage of the automatic squelch circuit described here is the fact that you do not have to be an expert to install it into the audio section of the receiver. Block diagram of the rectifier stage can be adjusted by means of preset potentiometer P3. The circuitry around electronic switch ES4 not only acts as a Schmitt trigger, but also ensures that the switch is not continuously opening and closing. When the voltage across capacitor CIO exceeds a certain value, ES4 is activated and the full supply voltage appears across resistor R13. The combination D2, RIO, R12 and C11 slows down the switch when this voltage changes value, thereby preventing short noise pulses from influencing the circuit. The junction of ES4 and R13 is connected to ES2 and ES3. The combination ES3 and R14 functions as an inverter and drives ESI. Thus, the circuit shown in the block diagram is realised. Switch ESI will be closed and ES2 will be open when only a little noise is detected, therefore the output of buffer amplifier A1 is fed to the input of the receiver audio stages. On the other hand, when a lot of noise is present, ESI will open and ES2 will close. Consequently, the loudspeaker will remain silent. Construction and calibration The printed circuit board and com- ponent overlay for the automatic squelch control is given in figure 3. As the circuit is relatively straightforward, construction should not present any problems. The same holds true for the installation; the volume control is quite easy to find and there is normally suf- ficient room inside the equipment to install the board. If not, the squelch circuit can be mounted in a separate small box. The supply voltage for the squelch circuit must be between 6 V and 12 V. The current consumption is only a few milliamps, therefore the receiver power supply can most probably be used. Calibration of the circuit is very straight- forward. The input level to A2 is preset by means of P2 in such a way that the noise peaks at the output of this opamp are correctly limited. The trigger thres- hold of ES4 (the lowest noise level at which the squelch circuit is activated) is set by means of P3. The setting of P2, although sounding complicated, is really quite simple. An incorrect setting of P2 means that the circuit switches on and off continuously. In which case P2 should be adjusted until the circuit reacts as it should. The automatic squelch control could be used in a number of applications such as CB transceivers, the MW receiver (Elektor March 1981) and the induction loop paging system (Elektor January 1982) when used as a 'babyphone' or intercom. H 3-28 - elektor i 1982 the DNR prir die MR printed cireuit board a practical noise reduction system Last month we promised to come up with a practical noise reduction system that avoids using a 'hard- to-get' 1C — and here it is. The circuit literally makes noise go 'off the air'. In addition to the usual HiFi applications, the circuit can be used to 'brush up' the sound quality of old records. They no longer need to be discarded because of the grating noise that made listening to favourites of years gone by almost unbearable. The same goes for FM radio: remote stations will sound much clearer once noise is eliminated. Noise is a universal problem, whether on television, radio, records or cassettes. It is even more irritating than distortion, especially in cases where the trebles are reproduced as piercing notes. As a rule, therefore, it is more important to accomplish a signal to noise ratio of 70 dB than a distortion level of —70 dB. This explains why there are so many noise reduction designs on the market, two of which, CX and DNR, were described in the last issue. This month we see how the DNR system can be put into practice. Like any other noise reduction system, the DNR circuit cannot be expected to work miracles. It makes the 'best of a bad job', for the only alternatives to noise reduction are to use a relatively noise-free signal source together with high quality equipment having a high signal-to-noise ratio. Let's face it, even high quality tuners using rotational, multi-unit aerials and professional tape recorders are not totally noise free. But, at least they do reduce noise to an acceptable level. It is when the use of less-than-top quality cassette recorders and gramophone records is contem- plated that noise reduction systems really can make an impressive improve- ment to the overall signal-to-noise ratio. As readers will remember, the DNR circuit described last month contained an 1C, the LM1894, which unfortu- nately is very difficult to obtain. The circuit in figure 1 gets around that problem by providing a substitute for the 1C, but at the same time it creates another snag: the circuit is not nearly as compact. Nevertheless, the board has been kept to a reasonable size and can be connected into a stereo system without any difficulty. The circuit Most of the circuit in figure 1 looks similar to a LM 1894 National Semicon- ductor application. IC1, a double OTA with darlington buffers, is in the centre. Two low-pass filters are built around the 1C and have a turnover frequency that Is dependent on the control current through pins 1 and 6. The greater the current flow, the higher the turnover frequency. The filter configuration is slightly different from the version shown in figure 5 in the February article. This time, the negative input of the OTAs (virtual ground) is driven instead of the positive input. The cur- rent source controlled capacitors C3 and C4 replace the active integrator. A capacitor voltage buffered by darlington transistors constitutes the output volt- age of the DNR circuit and this is reverse fed back to the negative input of the double OTA by way of R13 and R14. By way of a series resistor (R9 . . . R12), both OTA inputs are provided with a current which serves to improve the linearity of the input stage. After all, the OTA is simply a differential stage with a collector current that is equal to half the control current IaBC- Differen- tial stages tend to get overdriven rather easily, which is why the OTA input is often derived after a considerable volt- age division. As a matter of fact this is not necessary in this particular appli- cation, as the OTA used here is not the type to be overdriven. The circuit diagram for the dynamic noise filter substitute is very straightforward in- deed: it includes an RC filter with a resistor between its input and output and a capacitor between its output and ground. The dynamic constituent of the filter is provided by the variable RC time - in other words, the adjustable turnover frequency. The further the signal is filtered, the higher the voltage across resistor R. In figure 1 this will be seen to correspond to the current passing through C3 and C4, respect- ively. IABC determines the maximum current level. The filter attains its opti- mum performance when the turnover frequency is at a minimum (at around 800 Hz). This occurs when there is plenty of noise, but no other input signal to speak of. As soon as this happens, the IaBC (as will be apparent later) and, therefore the modulation, increases. Thus, the OTA operation is based on an increasing bandwidth to rising modulation ratio. Now to get back to the DNR circuit inputs. The emitter followers, T1 and T2, buffer the left-hand and the right- hand input signals, and for a very good reason. For not only does this provide the circuit with an input impedance of around 100 k, but the signals have to be buffered if the stereo pilot tone filter is to be added (between A and B and A' and B'). The filter must be driven from a source impedance of 4k7 and be terminated with the same value (R34 . . . R37). The filter may be necess- ary to make sure the pilot tone residues (19 kHz and 38 kHz) are below the noise level. What is at stake here is the effect of the pilot tone residues on the control loop, rather than on the output signal. Now that we're on the subject, let's take a look at the control loop. Resistors R5 and R6 add up the left and right channel input signals. The capaci- tors, C8 and Cl 9, serve to attenuate frequencies above 16 kHz. The wiper position of PI exerts a considerable in- fluence on the gain factor of the control loop. The latter determines the extent to which the L + R signal affects the turnover frequency of the two noise filters with the aid of the control cur- rent lABC- The circuit around A1 ampli- fies the control signal. Its gain factor is frequency-dependent. At very low fre- quencies, the gain of A1 is 4’/a; at fre- quencies above 6 kHz this rises to 100. The time constant formed by R24 and Cl 1 corresponds to a turnover fre- quency of around 6 kHz. A1 is followed 3-30 - eleklor march 1982 DNR printed I Parts List for figures 1 and 2 Resistors: R1,R2,R17,R18,R26,R27 = 100 k R3,R4,R15,R16,R24,R29 = 3k3 R5,R6,R7,R8,R13.R14 - 22 k R9.R10 = 56 k R1 1.R12 = 5k6 R19 = 15k R20,R23,R25,R33 = 10 k R21 = 330 k R22 = 82 k R28 -27 n R30 - 1 k R31 - 100 n R32- ion R34-.R35* -4k7 R36-.R37* - 6k8 PI = 100 k preset (see text) Capacitors: C1,C2 = 220 n MKH C3.C4 = 4n7 MKH C5,C6 • 4p7/16 V tantalum C7 = 10 p/16 V C8.C9 = 1 n MKH C10.C11 - 10 n MKH C12,C15,C17,C18 = 100 n MKH C13 = 6p8/16 V tantalum (or 4p7//2(i2) C14 - 33 n MKH C16 = 470p/25 V C19 = 220 p Semiconductors: T1,T2,T3= BC547B D1.D2- 1N4148 D3,D4,D5,D6 = 1N4001 IC1 = LM 13600 (National), Technomatic Ltd. IC2- LM 387 (National) IC3 = 78L12 Miscellaneous: Trl - 15 V/50 . ..100 mA transformer FI =31 5 mA fuse SI - mains switch • (see text) Instead of wire links A-B/A'-B', R34 . . . R37 and a single Toko pilot tone filter, type BLR3107N (FI1) may be connected. by the negative peak rectifier around A2. The storage capacitor C13 is charged from T3 by way of R28, provided the output voltage of A2 is sufficiently positive with respect to the voltage across C13 to make D1 conduct. As soon as this happens, the gain of A2 — in other words, the ratio of the emitter voltage of T3 to the output voltage of A1 — will be determined by the ratio of R33 to R30 and Cl 4 con- nected in series. Again, operation is based on frequency -dependent behav- iour. The control loop has the fre- quency characteristics of a high-pass filter featuring a turnover frequency of 6 kHz and a filter slope of 12 dB per octave. The reason for this parameter was explained in the February issue. By connecting R31 and D2 in series, the output of A2 is prevented from becoming too low when D1 no longer conducts. R32 and Cl 5 are also con- nected in series, which is necessary to limit the open loop gain of A2 during the periods that D2 conducts, whereas D1 does not. This is essential, since A2 (one half of the LM387) is compen- sated for a greater closed loop gain than while 02 conducts. The OTA control current 2 I ABC is determined by the voltage across Cl 3 and R29. The greater the voltage across Cl 3, the greater the control current and therefore the turnover frequency of the dynamic filters. The voltage across Cl 3 in turn depends on the level of the control signal; in other words, on the extent to which frequencies above 6 kHz are represented in the control signal. That covers the function of the control loop. A slight current I ABC passes through resistors R26 and R27. This is partly used to adjust the DC level of A2 (by way of R25). Something should be said about PI. This adjusts the gain of the control loop. The lower the wiper position of PI, the the DNR printed I arch 1982 - 3-31 Figure 3. How to connect the DNR circuit to the stereo in cases where recordings do not hive to be monitored during playback. Figure 4. Here the DNR circuit is connected permanently and exclusively to the playback module in the cassette deck. Figure 5. The most universal solution: the DNR can be switched off both for all types of signal greater the noise reduction. PI can be positioned in three different settings: 1 . The wiper voltage of PI is too low. This means not enough control volt- age is available, so that not only noise is reduced but so are the trebles. 2. The centre position. The noise re- duction is satisfactory without loss of trebles. 3. The wiper voltage of PI is too high, resulting in plenty of trebles and plenty of noise. The best setting for PI is half-way between 1. and 2. The DNR control can be switched off (the full bandwidth of 30 kHz, at least) by grounding the junc- tion of R30 and Cl 4. As a result, the control voltage is unable to reach the rectifier. In addition, the emitter voltage of T3 will be about 1 1 V, causing a high I ABC control current and therefore a high turnover frequency in the dynamic noise filters. In practice The printed circuit board for the DNR circuit is shown together with the parts list in figure 2. There is room on the board for a power supply, apart from the transformer, mains switch and fuse. It is equally feasible to connect a DC voltage of 15 V, provided the circuit is fed with a stabilised voltage neither above nor below 12 V. If the circuit is (also) to be used to reduce noise on FM radio, it may be necessary to include the pilot tone filter FI1 and the resistors R34 . . . R37. This depends on the pilot tone sup- pression capabilities of the tuner. The 19 kHz and 38 kHz pilot tone residues must be below the noise level. There are various ways in which to con- nect the DNR circuit to stereo equip- ment. Figure 3 makes use of the tape signal recording and playback facilities. These are available in practically any amplifier. The DNR circuit can be switched on and off with the monitor switch. It is no longer possible, however, to monitor recordings. Furthermore, the reserve inputs ('Aux') have to be used for playback purposes. One solution, according to the circuit in figure 3, is to switch the DNR permanently to play- back. In other words, the unit is not available for other signal sources. The most universal remedy is shown in figure 5, but this involves modifying the amplifier. The Elektor DNR prototype was tested thoroughly. All sorts of signal sources with various levels of noise were con- nected up. On the whole, the results were satisfactory. The setting of PI (noise reduction without loss of trebles) proved to be rather dependent on the signal source. It might be a good idea to substitute the preset for an ordinary potentiometer, but then again this depends on what the circuit is used for. At excessive noise levels during breaks in the music, audible fluctuations oc- curred in the noise volume. Again, this depends on the programme material. M 3-32 - ele ektor march 1982 aad acid battery charger lead add battery charger safe and easy to use Although NiCad batteries are relatively cheap, they by no means eliminate hermetically sealed lead acid batteries. For one thing, it is more economical to use them for high current consumption applications. As opposed to NiCads they are easy to charge, because they have a specific charge density. In addition, they can be connected in parallel to the load and a power supply and put into continuous operation. The circuit not only charges lead acid batteries, but also acts as a power supply. It is polarity-protected and includes current and voltage limiting. It also provides charge control and a polarity indicator. In other words, the battery charger is practically fool-proof I Compact lead acid storage batteries, like the well-known 'Dryfit' from Sonnenschein and YUASA from Japan are very popular with model hobbyists. Very often the smaller types (6 V and 12V; 1.1 Ah) fit in equipment that is normally supplied with baby or mono cells, for example, portable TV sets, video recorders and battery-powered cassette recorders. In such instances these batteries are a cost-saving alterna- tive to non-rechargeable batteries. Com- pared to the NiCad batteries they are very easy to recharge since they can remain inside the equipment. The power supply/charger is simply connected to the power supply socket of the device. It then takes over from the mains power supply while simultaneously recharging the batteries. As soon as the batteries are fully charged, they are topped up with a small 'stand-by' current. The charger may remain connected to the device for an unlimited period of The moment the mains plug is pulled out, the batteries automatically power the device, since they are permanently connected. The equipment merely has to be connected to the mains again for the batteries to be recharged. The printed circuit board for the lead acid battery charger is designed to accommodate various versions, with one or two minor modifications in component values. A choice may be made between an output voltage of 6 V with a maximum charge current of either 1 or 3 A and an output voltage of 12 V, again with either 1 or 3 A charge current. The charger is well protected against major disasters, such as short circuits, wrong polarity and/or power supply failure. It is almost im- possible to damage either the batteries or the charger. To make things easier, a LED is provided which lights up when the battery is connected the wrong way round. A second LED illuminates when the charge current starts to flow and goes out when this drops below a cer- tain level (the battery is fully charged) or in the event of a short circuit. One of the main advantages of the cir- cuit is its size. In spite of its com- pactness the printed circuit board has ample space for all the components. The charger board plus components cost much less than a ready-made charger. Lead acid batteries vs. NiCads Despite the fact that rechargeable dry batteries have improved in recent years and do not cause pollution like their NiCad counterparts, this form of power supply is steadily losing popularity. One of the main reasons for this is that dry lead acid batteries start to be avail- able from a nominal capacitance of 1 Ah, NiCads on the other hand can be acquired at much lower values. The ! ninimum nominal capacitance of 1 Ah oughly corresponds to a single cell (the ound cell by General Electric), whereas 5onnenschein offers series that can be ubstituted for 4 or 6 baby or single :ells. Compared to NiCads, lead acid batteries lave the following advantages: » The cell voltage is 66% higher, since it has a nominal value of 2 V. • It is very straightforward to control with the aid of a specific 'full' charge voltage. ■ They react better to both high and low temperatures. 1 They present a very low discharge. In fact they still rate 50% of their nom- tal capacitance after 16 months' quiesc- nce at 20°C. There is no danger of damage and loss of capacitance caused by changing ie polarity when the batteries are over- ischarged. c is not wise to store over-discharged atteries for more than four weeks lithout being recharged. Their current carrying capacity is very good. Dryfit batteries, for instance, that have a 1.1 Ah rating, can be loaded with up to 40 A. Very few NiCad types can beat that. But what about the disadvantages? As already stated before, low values are not available. Apart from this they have a few other drawbacks. Their lifespan is relatively short and they cannot always be charged rapidly. NiCad manu- facturers claim their batteries provide 500 times the nominal capacitance; whereas Sonnenschein types can only manage a factor of 200 and up to 1000 charging cycles in the case of partial discharge. This may sound a lot less, but in practice such batteries enable model enthusiasts to run a boat for several years. Even readers who are not interested in leak-proof lead acid batteries, may find a useful occupation for the charger described here. The circuit will act as an efficient 6 V or 12 V power supply and can also be used to charge car batteries. The circuit The circuit charges sealed lead acid batteries in a very straightforward manner. Readers merely have to keep an eye on the charge voltage and make sure this does not exceed 2.3 V per cell, to prevent an overcharge. Contrary to NiCad batteries, the initial charge state (partial discharge) is totally irrelevant. As a power supply therefore, the circuit is fully stabilised. In addition, the charge current should be limited too, so as to avoid an overload condition, since it can cope quite easily with high initial charge currents. The lead acid battery charger circuit is centred around the indispensable 723 1C. This meets the precisely calibrated output voltage and current limiting requirements. The trouble is, the 1C will not survive if a battery is connected with the wrong polarity, so that the circuit must include some form of charge current and po- larity indication. Figure 1 shows the result. As opposed to the standard 723 circuit shown in figure 2, the version in figure 1 uses fewer pin connections but more exter- nal components. These measures had to be taken to protect the 1C against nega- tive voltages in the event of an incor- rectly connected battery. Obviously, the fewer pins there are to protect, the easier it is to shield the 1C. The 723 now merely acts as a reference voltage source and transistors T1 . . . T5 constitute the opamp, the output stage and the current limiter. The voltage divider R1,R2 divides the nominal reference voltage of 7.15 V at pin 6 down to 6 V at pin 5. This enables an output voltage of 6.9 V to be im- plemented for the 6 V circuit. Pin 5 is the non-inverting input of the opamp inside the 723. The output voltage is fed back by way of the voltage divider R10, PI and R11 to the inverting input of the opamp (pin 4). Capacitor C3 serves to prevent oscillation and is con- nected between pin 4 and the output of the opamp at pin 13. Diodes D7 and D8 protect the circuit against polarity confusion by limiting the negative voltage to 0.7 V. The darlington output stage consists of T1 . . . T3 and provides the necessary current amplification. T3, a 2N3055, is well equipped to cope with the amount of dissipation expected due to the difference in voltage level between the non-calibrated voltage at the charge capacitor Cl and the output voltage (and current). T4 limits the output current. As soon as the voltage at the 'current sensor resistors', R4 and R5, drops to about 0.6 V, T4 starts to conduct and draws base drive current from T1. This stops the output current from rising any further. T5 is connected in parallel to T4. Nor- mally speaking, T5 does not conduct since its base voltage does not get a chance to become more positive than that at the emitter. This situation will only alter if a battery is connected with the wrong polarity. D9 will now be forward biased, enabling the transistor to be supplied with base drive current by way of R7. The transistor starts to Figure 2. By way of comparison! The standard power supply circuit using a 723 1C. conduct and practically 'shorts' the base emitter voltage produced by T1 . . . T3. The latter transistors will therefore be unable to conduct, which is the object of the exercise, for now no current can flow through this section of the circuit. Without this measure the battery would 'short' by way of D5 (or by way of D2 and D3 if D5 is not included). D5 protects pin 12 of the 723 1C. Now for the indicator section of the circuit. LEO 012 is usually 'off'. It will only light up, if the positive terminal at the output and the negative terminal are inverted. This happens if the battery is connected incorrectly. On the other hand, Dll is included in the collector circuit around T6. It lights as soon as T6 conducts, which occurs whenever the voltage at R8 drops to the level of the base emitter threshold vol- tage (about 0.6 V). Since R8 has the relatively high value of 56 fi, the vol- tage level concerned will be reached at an output current as low as 10 mA. Thus, D1 1 is an excellent means of controlling the charge current: it lights the moment a nominal charge current starts to flow. Diode D6 is connected in parallel to R8 to allow the charge m JTJ y-O-O , m/ ljH IlDflOOll) R1 = 680 n R2 = 3k3 R3 = 2k2 R4.R5 = 1 n/0.5 W, for 3 A: 0.33 fl/1 W R6 = 22 k R7 = 4k7 (10 k) R8 = 56 SI R9= 100 SI R10 - wire link (6k8) R11 = 4k7 R12 = 470 SI (680 S2) PI = 2k5 preset Capacitors: Cl = 1000 p/16 V (25 V), for 3 A: 2200 (i/16 V (25 V) C2.C4- lOOn C3- 10 n Semiconductors: D1 . . . D6- 1N4001 for 3 A: 1N5401 D7 ... DIO - 1N4148 D11.D12- LED T1.T4.T5 = BC547B T2 - BD 135, BD 137, BD139 T3 - 2N3055, for 1 2 V/3 A : 2 x 2N3055 T6-BC557B IC1 - 723 10V/1.5 A sec. 18 V/1.5 A sec. 10 V/5 A sec. 18 V/5 A sec. 3-35 elektor march 1982 - current to rise above 10 mA, if necess- ary. LED Dll is lit, provided a charge current flows through the circuit, the battery polarity is correct and no short circuit is produced at the output. 1 A operation The circuit can be constructed for either 6 V or 12 V. The component values required for the 12 V version are indicated in brackets in the circuit diagram and the parts list. Apart from I the transformer and the electrolytic capacitor only three resistors (R7, RIO and R12) have to be modified, if the | 12V version is chosen. Where 6 V batteries have to be charged, the output voltage is adjusted with PI to 6.9 V (±0.1 V) when the circuit is quiescent. This can be done with a multimeter. In the case of 12 V batter- ies, the quiescent voltage of the charger must be adjusted to 13.8 V (±0.1 V). Transistor T3 must always be cooled. In 1 A applications, however, the heat sink can be relatively small and can even be omitted if the transistor is mounted on the back of a metal case. 3 A operation The above also applies to 6 V/3 A and 1 2 V/3 A circuits, only now the trans- former, capacitor C2, diodes D1 . . . D6, R4 and R5 must be modified to cope with the higher output current. The new values are indicated in the parts list. In the 3 A output current circuits, cooling transistor T3 is a little more critical. At an output voltage of 6 V, a heat sink of 2°C per Watt will guarantee enough heat dissipation even if a short circuit con- dition lasts for a relatively long time. At 12V the transistors have to dissipate a considerable amount of power. In the case of a short circuit, T3 has to get rid of some 50 Watt. Provided the short circuit does not last longer than a few minutes, a heat sink may be used with a heat resistance of 1 .5°C per Watt. If the circuit is to be short proof for longer periods, however, it is advisable to distribute the output power between two transistors, as shown in figure 3. Charging car batteries The 3 A version is particularly suitable for charging car batteries. About 36 Ah can be charged during the night. Using the indicated output voltages of either 6.9 V or 13.8 V, starter batteries can be recharged to about 75% of their nom- inal capacitance. Generally, this should be enough to revive a dead battery. Furthermore, the battery can be con- nected for an unlimited length of time at these voltage levels. Readers who intend to use the charger for this purpose only should set the output vol- tage at a higher value to be on the safe side. If 2.4 V is provided per cell, the battery will reach 80% of its nominal value and 2.65 V will bring it up to 100%. Once the battery is fully charged, topping it up any further will damage it in the long run. If the battery is to be charged overnight, an output voltage of either 8 V or 16 V will be perfectly safe, but do not forget to disconnect the charger in the morning! With regard to R4 and R5, they may be replaced by a single resistor that has half the value but double the load capacity, such as 0.47 WVI for 1 A, for instance. The LEDs may be any colour, since it makes no difference to the circuit. In the prototype the charge control LED (Dll) was green and the polarity indicator (D12) was red. M illy charged. 3-36 - elektc blue LEDs Silicon carbide is by no means a 'new' been carried out to date is rather semiconductor material, even though it limited. Furthermore, scientists have as has come into vogue only fairly recently, yet failed to come up with a practical In fact it is one of the oldest materials, method for processing single-crystal its electroluminescence being reported silicon carbide, which is essential if the by Round as early as 1907 (Round was semiconductor material is to be im- working with SiC crystals at that time), plemented in electronics. As far as its semiconductor properties Consequently, semiconductors were in- are concerned, SiC is similar to silicon, itially made of germanium and later but there are several essential differ- silicon, using increasingly advanced ences. SiC has a non-axial crystal struc- technology. It is only now, when the ture, a large unit cell and a large band sky seems to be the limit as far as silicon gap, meaning that the physical phenom- applications are concerned, that less ena observed are extremely complicated common semiconductor materials, such to interpret. as gallium, arsenic and silicon carbide, Unlike other large-gap semiconductors, are being rediscovered. This is because SiC can easily be doped both p- and n- there are a few areas in electronics to type, although involved techniques have which they are particularly suited, to be developed to deal with its extreme Gallium, for instance, is ideal in LEDs hardness and chemical inertness. For this and RF semiconductors. Now that reason the amount of research that has silicon carbide has been found to emit blue light, the 'file' dating back to 1907 blue LEDs has as last been reopened. But before we examine the properties of SiC in detail, let us find out how semiconductors emit light in general. silicon carbide may provide the answer Large-gap semiconductors are potentially useful materials for the manufacture of light-emitting diodes. Their spectral range now includes the blue and ultra-violet regions. In addition, some of the materials can be used to manufacture high-power microwave devices and sensors for high-temperature operation. During the past three years, the physical properties of SiC (silicon carbide) have been the subject of an enlightening study. This article describes this semiconductor material and shows how blue-emitting SiC diodes can be produced by methods similar to those for existing GaAs (gallium arsenide) or GaP (gallium phosphide) devices. Semiconductor light Any semiconductor will emit light at a certain temperature. The material goes dark red in the 700 . . . 900°C range and literally white hot at higher temperature levels. The semiconductor will then behave in the same way as a light bulb or even the flame of a candle. Due to their luminescence, however, semicon- ductors also emit light at much lower temperatures. The term 'luminescence' was introduced by Wiedemann in 1889 to denote any form of light emission that is not caused by the temperature of the light-emitting material. It is a common phenomenon and can be seen in fluorescent TV screens, etc. Light emissions are based on the follow- 3 ing principle. When an atom is supplied with energy, it is stimulated and absorbs the energy. An atom can only be energized very briefly before returning to its stable ground state. The absorbed energy is then released as electro- magnetic radiation which assumes the form of visible light when it coincides with a certain wavelength. Bohr's atomic model as illustrated in figure 2 , can be used to visualise this process: atoms move in a fixed orbit around the nucleus, rather like planets around the sun. Energy in the form of a high-speed electron is propelled in from the exterior and collides with one of the electrons belonging to an atom. This absorbs the incoming energy and is launched into a higher, more powerful orbit. The whole process lasts a very short time, after which the electron returns to its original position while releasing its surplus energy. The wave- length of the emitted radiation depends on the difference between the energized and the non-energized state. In the 380 . . . 750 nm range (see figure 1) the radiation will be visible as light. Atoms can be stimulated in other ways as well, with the aid of X-rays, light, particle radiation, or heat, for instance. The same principle applies to lumi- nescence in semiconductor materials. Again, light is produced by electrons returning from a high energy to a low energy state while releasing their excess energy, usually in the form of heat (phonon vibration), but sometimes as radiation (photons) in the infrared and visible light range. The charge carrier energy dissipation process described above occurs in the polarised pn junction of a diode that is forward biased. To understand this process, let us make a short 'excursion' into semiconductor technology. In semiconductor materials, electrons assume certain levels of energy only. The valence band and the conduction band both have the highest energy levels for electrons in normal semiconductor materials. The separation between the top of the valence band and the bottom of the conduction band is known as the energy gap and is shown in figure 3. If the semiconductor material is pure, electrons can not exist in this 'forbidden' gap. Electronic states are produced in the gap by introducing impurities. The maximum energy level of the emitted photons is determined by the band gap energy of the solid in which the pn junc- tion is formed. Suitable materials for LED devices are GaAs, GaP and SiC. Thus, blue light having a wavelength of 380 . . . 440 nm in the short wave region of the emission spectrum can only be derived from semiconductor materials that have a corresponding band gap. This is why gallium junctions, for instance, cannot emit blue light. Table 1 provides a survey of various semiconductor materials and lists them according to their band gap, wavelength (if available) and radiation range. Semiconductor photo diodes Figure 4 shows the structure of a semi- conductor photo diode. It consists of n doped and p doped semiconductor material. The area between the p zone and the n zone is called the boundary layer or junction, in which the illumi- nating recombination occurs. The doping material in the p zone contains atoms which ail have one valence elec- tron less than semiconductor material. 3-38-' blue LEDs Photo 1 . X-ray of a silicon carbide wafer which can be used in spite of the irregularities shown. 4 forward direction (positive pole of the battery at the p side, negative pole at the n side), electrons and holes are injected into the boundary layer. Now holes belonging to the p side reach the n zone and 'recombine' with the abundant free electrons. Similarly, electrons are sent from the n side to the p zone where they also recombine. A distinction is made between direct recombination (where an electron is moved directly from the conduction band into a hole in the valence band) and indirect recombination (where the recombination is not carried out directly between the bands, but between the bands and the transition levels between them). This is shown in figure 5. The most favourable ratios are obtained using 'direct' semiconductors (able to be recombined directly), which emit light provided the band gap width is sufficient. Indirect semiconductors are also able to emit light at a certain band gap width. This can be controlled by injecting foreign atoms, 'iso-electronic centres'. GaP LEDs, for instance, are doped with nitrogen to make them emit green light. Injecting zinc oxide, on the other hand, causes them to emit red light. OH mu 0© 0© 0© 0© 0® t-* 0® 0®