ISSN 0970-3993 73 MAY 1989 THE MAGAZINE FOR PROFESSIONALS TELECOM, A MUST IN ELECTION MANIFESTO THE BUGGING STORY OF INFORMATION VIDEO RECOR- DING AMPLIFIER DYNAMIC RANGE PROCESSOR MULTI-POINT IR CONTROL CLASS-D AMPLIFIER mi m 'mmm imm If your child is a CHIP of the OLD BLOCK, He deserves / | F| The Electronic Comic-Books in Electro Technology From the publishers of ©Isktor RESI & TRANSI I BANISH THE MYSTERIES OF ELECTRONICS! In the course of the book, they explain how to build a few practical circuits, a continuity tester, a morse code, bleepes & an amplifier. A Printed Circuit Board is included to simplify the actual construction as well as an extremely useful gadget, The Resimeter RESI & TRANSI II Hands of my Bike!! Electronics & Cartoons? Why not? In this, their Second Book Resi & Transi set about building an Electronics Alarm with a powerful siren to protect a bike, motorbike, car or even a house. And, as usual they are totally indifferent to the normal code of practice of electronic components. But then, who ever said the electronics has to be dull? Resi & Transi make ELECTRONICS easy .... and fun. Send full payment by M.O./I.P.O./D.D. No Cheque Please, to: - precious — ELECTRONICS CORPORATION 52-C, Proctor Road, Bombay-400 007 Ph. 367459, 369478 Telex-(OII) 76661 ELEKIN Publisher: C.R. Chandarana Editor: Surendra Iyer Technical Adviser : Ashok Dongre Circulation: Advertising: J. Dhas Production: C.N. Mithagari Address : ELEKTOR ELECTRONICS PVT. LTD. 52. C Proctor Road, Bombay-400 007 INDIA Telex: (011) 76661 ELEK IN Volume-7 Number-5 May-1989 CONTENTS Overseas editions: Elektor Electronics 1, Harlequin Avenue. Great West Road, Brentford TW8, SEW U.K. Editor: Len Seymour Pulitron Publicacoes Tecnicas Ltda Av Ipiranga 1100, 9° andar CEPkOKMO Sao Paulo - Brazil Editor: Juliano Barsali Elektor sari Route Nationale; Le Seau; B.P. 53 592270 Bailleul - France Editors : DRS Meyer; G C P Raeder9dorf Elektor VerlBg GmbH Susterfeld-StraBe 25 100 Aachen - West Germany Editor : E J A Krempelsauer Elektor EPE Karalskaki 14 16673 Voula - Athens - Greece Editor: E Xanthoulls Peter Treckpoelstraat 2.4 6191 VK Beek - the Netherlands Editor: PEL Kersemakers Ferreira & Bento Lda R.D. Estefania, 32,1° 1000 Lisboa - Portugal Editor: Jorge Goncalves Ingelek S.A. Plaza Republica Ecuador 2.28016 Madrid-Spain Editor : A M Ferrer In Part : Kedhorn Holdings PTY Ltd Cnr Fax Valley Road & Kiogle Street Wahroonga NSW 2076 - Australia Editor: Roger Harrison Electronic Press AB Box 63 182 11 Danderyd - Swceden Editor: Bill Cedrum The Circuits are domestic use only. The submission of designs or articles implies permission to the publisher to alter end translate the text and design and to use the contents in other Elektor Publications and activities. The publishers cannot guarantee to return any matorial submitted to them. Material must be sent to the Holland address {given above). All drawings, photographs, printed circuit boards and articles published in elektor publications are copyright and may not be reproduced or imitated in whole or part without prior written permission of the publishers. Patent protection may exist in respect of circuit devices, components etc. described in this magazine. The publishers do not accept responsibility for failing to identify such patent or other protection. MEMBER Printed at : Trupti Offset Bombay - 400 013 Ph. 4923261,4921354 Copyright ® 1988 Elektuur B.V. Editorial Optical Computing 5.07 Special Features Telecom, a must in election manifesto 5.10 The bugging story of information 5.16 Audio & Hi-Fi PROJECT: Class-0 amplifier 5.31 Horn loading revisited 5.53 Application notes Dynamic range processor type SSM 2120/2122 5.39 Components Practical filter design (4) •• 5.34 Design Ideas New circuit protection devices for loudspeaker systems 5.36 General Interest PROJECT: The digital model train part 3 5.26 PROJECT: Multi-point infra-red remote control 5.40 Radio & Television PROJECT: Video recording amplifier 5.38 PROJECT: High-precision DLF-based locked frequency reference 5.48 Science & Technology Big strides in molecular electronics research 5.46 Information Telecom News : 5.20 Industry News .T. 5.21 Electronic News 5.23 Guide-Lines Classified ads 5.74 Index of advertisers 5.74 elektor india may 1989 5.05 *_ Front cover People working in isolated areas of the world may be cut off completely from nor- mal means of com- munication. This portable system, which allows engineers on exploration work, rescue teams, businessmen, reporters, and others whose job takes them 'out in the field' to make clear contact by telephone, or to transmit and re- ceive data, has been developed by Marconi Marine. Called 'Satpax', the system links - via the satellites of the Inter- national Maritime Sat- ellite Organization - to the international tele- phone and telex net- work. Marconi claims that it can be used almost anywhere In the world. OPTICAL COMPUTING The news that Mitsubishi has succeeded in making the world's first optical chip must fill most of us with excitement Light, or rather photons, is, of course, already being used on a rapidly ex- panding scale in telecommunications (fibre optics) and in optical memories (compact disc), but not yet in computing, This is not for want of trying, for research into optical computing has been going on for some years in several parts of the world, notably in Japan, the United Kingdom and the United States. Optical computing is the science and technology of advanced computing for the future, because we are slowly nearing the final limitations of electronic computers. This is because the silicon chips used in current computers have a number of drawbacks: the speed at which electrons can travel through silicon is restricted to 5X10 5 m/s; also, it is already becoming more and more difficult to etch ever narrower paths into the silicon chips (which, incidentally, increase the risk of cross-talk); and finally, as they shrink, it becomes harder to keep them cold and operating properly. Until the Mitsubishi breakthrough, the only hardware developed consisted of (relatively) simple bistable optical devices (bods), also called transphasors. The Japanese have succeeded in developing a practical optical bonding tech- nology, so that the device they have made consists of many transphasors bonded three-dimensionally and constructed, with a number of other (elec- tronic) components on to a GaAs substrate measuring only 3x3 mm. If these initial reports are substantiated, it appears that the Japanese have suc- ceeded In producing the first building brick for a single-board optical computer. * * * INCREASE IN COVER PRICE Rising costs, particularly in news-gla?ed paper, forced us to reluctantly revise the cover price to Rs. 10.00 from this issue onwards. elektor India may 1989 5.07 THE DIGITAL MODEL TRAIN — PART 3 by T. Wigmore In Part 2 we described the background and design of a locomotive decoder and a two-rail adaptor. This month’s instalment is dedicated primarily to the construction of those units, but it will also describe a modification to the decoder involving a different decoder chip. As we have seen last month, the decoder is designed in surface mount technology (SMT). It is, however, impossible to ob- tain ah components as SMT types. Some ICs, for instance, depend for cooling on their housing and if this becomes too small, a heat sink has to be used, defeating the object of the exercise. Therefore, the decoder is a hybrid cir- cuit. It uses a double-sided PCB that is populated at one side with conventional components and at the other with SMT components. Soldering surface-mount components Working with surface-mount (SMT) components requires rather more dex- terity, patience and accuracy than work- ing w'ith conventional components. SMT components must be soldered direct to the printed-circuit board (PCB): IC holders, for instance, are no longer re- quired. Because of that, it is essential that during 1 soldering appropriate pre- cautions are taken to ensure the absence of any electrostatic charges (earthed working surface and soldering bit, for example). The soldering iron should be rated at about 15 — 18 W, have temperature con- trol and a fine-pointed bit. One of the first things that strikes one on a first acquaintance with passive SMT components is that most of them no longer show their value. This makes it essential not to remove them from their packing until they are really required. Apart from very thin solder, there is also a special solder dispenser on the market that is ideal for soldering SMT compo- nents. A new aid is solder in a syringe. This consists of small granules of lead- tin alloy suspended in a semi-liquid paste. When this paste is applied to the solder pad(s), the component terminal(s) may be pressed into the paste, after which it only requires a touch of the hot soldering bit to give a clean joint. 5.26 elektor indis may 1989 Soldering ICs is done in very much the same way. Note, however, that SMD ICs do not have the usual identification of pin 1. This is located at the most oblique side of the device (see also Fig. 20). Construction The ready-made printed-circuit board consists of four sections; two locomotive boards (double-sided, but not through- plated) and two two-rail adaptor boards (single-sided) as shown in Fig. 18a and 18b respectively. Construction should be started with cutting the PCB into these four sections. Determine the voltage for the head and tail lights. As stated in Part 2, the stan- dard voltage for the lights is 10 V, which is perfectly satisfactory for 12-V or 16-V bulbs. If the higher supply of 20 V is needed (because 24-V bulbs or two 12-V bulbs in series are used), a hole of 0.8 — 1 mm dia. must be drilled in the board for pin 9 of IC5. Also, in that case, the through-connection from one side of the board to the other, coinciding with pin 2 of ICj must NOT be made. Mount the (only) wire link to the left of ICi and the four through connections shown in Fig. 19a (with the aid of short lengths of equipment wire). Cut these wires as close as possible to the board, particularly where later ICi and ICs will be located, except beside pin 8 of ICi, which should protrude 3—4 mm at the SMT side. Select the wanted locomotive address with the aid of Table 3 and install this as shown in Fig. 19b. It is possible to change this afterwards but, owing to the presence of ICi, that will then be a tricky operation. Solder IC 2 and ICi at the SMT side of the board. Pins 8 of IC 2 and 2, 4 and 12 of IC3 coincide with a through connec- tion yet to be made. Bend these pins with a pair of small pliers so that they drop readily into the relevant holes. Before soldering, add. a short length of equip- ment wire that can be soldered at the non-SMT side as well (see Fig. 19c). Note that when a lamp voltage of 20 V is used, the through connection at pin 2 of ICj must NOT be made. Mount all components, except the two ICs and D9, at the non-SMT side of the board. Some conventional components need, none the less, be fitted as if they were SMT types: Di, D: (if a 1N4148 is used). Da and the anode of Dm. Bend the terminal wires of these components sharply near the body and cut them close to the body. If the two-rail adaptor is going to be used, the cathode of D? should protrude about 3 mm at the SMT side: later this will be used as through connection with the adaptor. Capacitors Cj and Cs (beware of the polarity) are bent at right angles over R* and Di re- spectively as shown in Fig. 19d. Circuits ICi and ICj must be soldered direct (no holder) to the board. A num- ber of pins must be shortened, because they must not protrude through the board: pins 1, 2, 3, 4, 9 and 11 of ICi, and pins 1, 7, 9 (not with 20 V lamp voltage), 11 and 14 of IC5. The other pins are soldered at both sides. Mount D 9 at the non-SMT side. Finally, solder all components at the SMT side of the board. Note that this side also has to house the four (non- SMT) free-wheeling diodes, Di to De. The terminals of these diodes, except the cathodes of Di and Dj, should be soldered direct to the board. il ij l = LOCOMOTIVE DECODER All parts surface mount assembly except when marked *. Resistors I ± 5%|: Rl = 12K R2;R4;R8 = 100K R3;R7 = 47K R5 = 270K Rb = 2K7 1...4 = through connections to be .'hade 5 = wire link 6 = address setting (see Fig. 19b) Capacitors: Cl = 3n3 C 2 = lOn C3 = 470p C4= 1 Ojj; 25 V: tantalum Ca=47/r; 6V3: tantalum Semiconductors: Di= 5V1; 400 mW zener diode * D2 = BAV100 D3. . .Da incl. = 1 N4 1 48 + Dz. . .Dio incl. = 1N4001 + ICi =MC145029 (Motorola) + IC2-4060 IC3=4585 IC4 = 4001 ICa=L293 (SGS) dr LM 18293 (National Semiconductor) + TWO-RAIL ADAPTOR All parts surface-mount assembly Resistors (±5%): Ri;R2 = 1M0 R3=10K R4 = 270K Capacitors: Cl = InO C2=15n C3=100p C4=47n Semiconductors: 1C 1=4030 IC2 = 4538 Miscellaneous: PCB Type 87291-2/3 (cut in 4 tor 2 two-rail adaptors and 2 locomotive decoders). Fig. 18. Layout and track side of the double- sided (not through-plated) decoder board (a) and those of the single-sided two-rail adaptor hoard (b). This concludes the construction of the locomotive decoder board. Construction of the single-sided two-rail adaptor board is straightforward. This board is populated with SMT compo- nents only. Before the two boards are sandwiched (if the adaptor board is used), they should be tested together. There are four elec- trical connections between them that. for test purposes, should be made before some of the wires are cut short. These are: B (made with the cathode terminal of D7 that has been kept long for this purpose); earth (made with the through connection adjacent to pin 8 of ICt); the positive supply line and the data line. The last two should be made with short lengths of equipment wire — see Fig. 19e. Fig. 19. How to make the through connec- tions (a), but note that only three are made in this photograph; (b) the address setting of pins 1 — 4 of ICi; (c) the through connec- tions of ICi and ICi to the other side of the board; (d) C4 and Cs arc bent at right angles across It* and Di respectively, while some other components, such as Di, are mounted as SMT elements; (e) the decoder and two- rail adaptor boards are sandwiched. elektor india may 1989 5.27 Installation The locomotive decoder can be used with d.c. as well as with a.c. locomotives. Actually, the motor of an a.c. loco- motive is generally connected as a d.c. motor. The way this is done is shown in Fig. 21. The general wiring diagram of an a.c. motor (here a Marklin type) is shown in Fig. 21a. It is seen that it is connected as a scries motor whose stator winding has a centre tap. Which half of the stator winding is used at any one time is determined by the position of the change-over relay (in some models this relay also switches over the lights). As already discussed, this change-over relay must be removed from Marklin locomotives. Figure 21b shows how two diodes in series with the disconnected terminals of the stator winding convert the a.c. motor to a d.c. motor. The direc- tion of rotation is then dependent on the polarity. The two motor terminals are connected to the motor output of the decoder (to the right of ICs). The decoupling com- ponents, Li, Ci, C 2 , may be retained, but C 2 , which was connected to earth must now be connected to the ‘0’ line of the decoder. D.C. motors can be connected to the motor terminals of the decoder without any modifications. There are various ways of connecting the lights and some of these are shown in Fig. 22. For instance, they may be switched in the positive line (22a) or in the negative line (22b). The former is preferable in view of the somewhat lower dissipation in IC5. If the lights are required to be indepen- dent of the direction of travel, the lamps are connected in series, two by two, direct to the rail voltage as in Fig. 22c. If the lights are preferred in parallel, but are not suitable for 20-V operation, they may be connected to the two La ter- minals (Fig. 22d). The voltage for the lights must then be set to 10 V as dis- cussed earlier. When Marklin locomotives are used, it is important that if the lights are connec- ted to the La terminals on the decoder there is no connection between them and the chassis of the locomotive. This means that either the lamp holders must be of the insulated type or the lamps must be connected via an additional di- ode as shown in Fig. 22e. A drawback of the diode solution is that the lights often have no constant brightness and may flicker from time to time. If Marklin locomotives are used and the lights are required to be operated in- dependently of the direction of travel, they should be connected as shown in Fig. 22f. They need not be isolated from the locomotive chasses. The brightness may be set by giving the series resistor an appropriate value. The motor must also be electrically isolated from the locomotive chassis, but 5.28 elektor India may 1989 Fig. 20. Pin 1 of an SMD IC is located at the most oblique side. 87»1-m- 10a Fig. 21. Circuit diagram of the Marklin motor connections (a); this a.c. motor may be converted for d.c. operation as shown in (b). Fig. 23. Decoder (without two-rail adaptor) installed in a Marklin locomotive; the buffer capacitor, w;here used, is fitted in the back of the locomotive. The arrow points to the shrink-sleeved diodes that enable the polarity of the stator field to be reversed. Fig. 22. Various possibilities of connecting the lights. Table 3. number of loco- motive address number of loco- motive address A1 A2 A3 A4 A1 A2 A3 A4 01 1 0 0 0 41 X 1 1 1 02 X 0 0 0 42 0 X 1 1 03 0 1 0 0 43 1 X 1 1 04 1 1 0 0 44 X X 1 1 05 X 1 0 0 45 0 0 X 1 06 0 X 0 0 46 1 0 X 1 07 1 X 0 0 47 X 0 X 1 08 X X 0 0 48 0 1 X 1 09 0 0 1 0 49 1 1 X 1 10 1 0 1 0 50 X 1 X 1 11 X 0 1 0 51 0 X X 1 12 0 1 1 0 52 1 X X 1 13 1 1 1 0 53 X X X 1 14 X 1 1 0 54 0 0 0 X 15 0 X 1 0 55 1 0 0 X 16 1 X 1 0 56 X 0 0 X 17 X X 1 0 57 0 1 0 X 18 0 0 X 0 58 1 1 0 X 19 1 0 X 0 59 X 1 0 X 20 X 0 X 0 60 0 X 0 X 21 0 1 X 0 61 1 X 0 X 22 1 1 X 0 62 X X 0 X 23 X 1 X 0 63 0 0 1 X 24 0 X X 0 64 1 0 1 X 25 1 X X 0 65 X 0 1 X 26 X X X 0 66 0 1 1 X 27 0 0 0 1 67 1 1 1 X 28 1 0 0 1 68 reserve 29 X 0 <9 1 69 0 X ...1 X 30 0 1 0 1 70 1 X 1 X 31 1 1 0 1 71 X X 1 X 32 X 1 0 1 72 0 0 X X 33 0 X 0 1 73 1 0 X X 34 1 X 0 1 74 X 0 X X 35 X X 0 1 75 0 1 X X 36 0 0 1 1 76 1 1 X X 37 1 0 1 1 77 X 1 X X 38 X 0 1 1 78 0 X X X 39 0 1 1 1 79 1 X X X 40 1 1 1 1 80 0 0 0 0 that is normally the case anyway. The above explanations should ensure that the conversion of d.c. motors for use on a Marklin system should not present any problems. Note, however, that the locomotive must be provided with a sliding contact. Connecting the decoder and two-rail adaptor to d.c. locomotives for two-rail systems should not present any difficult- ies. The B(rown) and R(ed) terminals may be connected to the wheel contacts at will, since the decoder is not polarity- conscious. Testing and faultfinding It is advisable to test the decoder in as- sociation with the locomotive before in- stalling it. In the following it is assumed that at least a Marklin digital HO system is available, that is, one Central Unit, one Control 80, and one 16-V mains transformer. Later in the series, this may also be the Elektor Electronics Digital Model Train System (EDiTS). Connect the brown and red wires of the HO system to the B and R terminals re- spectively on the decoder. Provided that the address keyed in on Control 80 cor- responds to the address set on the decoder, the locomotive should react to an adjustment of the speed regulator. If it does not, check that the locomotive address has been set correctly (Table 3) and that the supply for the logic circuits is 4.5— 5.5 V. If these are correct, make sure that pin 12 of IC 2 is logic 0 and that the oscil- lator operates (this is indicated by the logic level at pin 1 of IC 2 changing every second). If all these are in order, measure the average output voltage at pin 1 of ICs: it should be possible to vary this with the Control 80 speed regulator between 0 V and just below the level of the supply voltage to the logic circuits. If this is not possible, check whether the logic level at pin 5 of ICi changes when the function key on Control 80 is operated. If all these are correct, the fault lies in ICi (incorrect address or baud rate). If the voltage at pin 1 of ICs can be varied, but the locomotive does not move, check whether pin 2 of ICs is logic 0 and pin 7 of IC 5 is logic 1 (or the other way round, depending on the position of the function key). If this is in order, the fault lies in ICs, or the motor is connected incorrectly, or the motor is (mechanically) impeded. In this context, note that the outputs of ICs are not short-circuit proof: care should, therefore, be taken when lights and motor are connected. Thermal- overload protection is provided. If the decoder operates correctly, but the last received data are lost rapidly (in spite of the external buffer capacitor) when the supply is switched off (emergency control), too high a current to the logic circuits is indicated. A poss- ible cause of this is a short-circuit be- tween two logic outputs or between such an output and a supply line (check the non-used outputs of IC 2 as well), or a zener diode with a very high leakage cur- rent. Note that solder flux is electrically conductive and may, therefore, be responsible for a short-circuit. Another possible cause of too high a current may lie in the particular type of 4060 chip. Although in all makes the outputs are reset, in some of them the in- ternal oscillator is not switched off. Installation It is advisable to glue (two-component epoxy resin or super glue) a small rec- tangular piece of aluminium across the decoder (ICi-ICs side): this will act as a heat sink for ICs. This circuit can get pretty hot during driving, but that is normal. As long as you can put your finger to it without getting burnt, nothing untoward is happening. It can- not be too strongly emphasized to take care that there is no short-circuit be- tween the decoder and the locomotive chassis. The use of an external buffer capacitor depends on the circumstances. It is essential when tracks with conventional block protection are used, that is, those that stop the locomotive by removing the voltage from the rails. If the locomotive is stopped by switching the rails to a lower voltage (say, 8 V), an external buffer capacitor is not required. Although the charging current of the buffer capacitor is some 200 times higher than its discharge current, it may take a short time after the power is ap- plied to the track before (he decoders are ready for operation. This time lag depends on the. capacitance of the buffer, and it is, therefore, not advisable to take too large a value for this compo- nent — some thousands of microfarads is quite sufficient. Sidings and passing loops • A classical problem with two-rail tracks is the short-circuit between the two rails when a passing loop is used. In a con- ventional system, the loop is electrically insulated. The locomotive moves on to the loop and while it is there the polarity of the rail voltage is changed over. Un- fortunately, this also affects the direc- tion of travel of all other locomotives on the track. The digital system obviates this problem. The loop is again electrically insulated from the remainder of the track. The locomotive moves onto the loop as before, but in this case the polarity of the voltage on the loop is reversed while the locomotive is on the loop. This will not affect the direction of travel, because that is, after all, determined internally and independently of the polarity of the connections. Operational hints As discussed in Part 2, if the decoder is used in conjunction with the Marklin digital HO system, the function switch on Control 80 changes the direction of travel. One Control 80 suffices to control a number of locomotives, but as soon as a change-over from one locomotive ad- dress to another takes place, a problem arises. The function of the newly ad- dressed locomotive will be set automati- cally to ‘off’. If this locomotive was previously set to ‘reverse’, it will sudden- ly change direction. This is prevented by keying in only the first digit of the new address on Control 80, then pressing the function switch (only if the relevant locomotive is set to ‘reverse’, of course), setting the speed controller to the pos- ition it was when the locomotive to be addressed was last controlled, and only then keying in the second digit of the ad- dress. elektor india may 1989 5.29 Decoder modification When the design of the decoder was started just about a year ago, availability of the MC145029 chip was assured by Motorola Europe. In spite of that assurance, production of the device was stopped in December 1988. Many con- structors will, therefore, find it next to impossible to obtain the chip. For that reason, a modification was designed that is based on the MC145027 (which is also used in the points/signals decoder — see Part 1 — and is, incidentally somewhat cheaper than the MC145029). In the MC145027, the first five bits are address bits and the other four are data bits, while in the locomotive decoder four address bits and five data bits are required. There is thus a shortage of one data bit: the bit that ensures changeover of direction of travel. Fig. 24. The decoder board modified to oper- ate with an MC145027 data decoder. It is thus necessary to effect the change of direction of travel by another means, and this is done by adding a dual J-K bistable (physically glued on top of the MC145027 as shown in Fig. 24 and Fig. 27). This modification makes the change-over of direction of travel com- patible with the Marklin system — see Fig. 25. Fig. 25. The modification restores the change of direction of travel to the original Marklin concept. Parts required for modification R9 = 1 MO (physically as small as possible! D13,D14,D15 = 1N4148 1C 1= MCI 46027 (in place of MC145029) IC6 = 4027 (SMD) logic Fig. 26. Circuit diagram of the modification. The lowest speed-step is decoded and used to clock the switching bistable. This is achieved as shown in Fig. 26. The unused bistable in the SMT 4027, FFi, is used for decoding the lowest speed-step. It is set when its input is 1000 after which is clocks FF 2 . Since the J and K inputs of FF 2 are strapped together, the logic level of the output of the bistable changes with every clock pulse. It is thus not possible to change the direction of travel two times in quick succession, but that does not occur in practice very often in any case. If, therefore, the direction of travel was changed erroneously, this can only be corrected after FFi has been reset and this does not happen until Di, Ds or D« Fig. 27. Adding three diodes and a dual bistable on to the MC 145027 requires some dexterity. has become logic 1, that is, until the locomotive has travelled a short dis- tance. The construction of the modification re- quires dexterity. The pins of the MC145027, like those of the MC145029, are cut short, but pin 5 is bent away from the body slightly to enable it being soldered to the earth track adjacent to the ICi position. As stated, glue the SMT 4027 on top of the MC145027 in such a way that pins 1 and 16 of the two ICs coincide. The anodes of diodes Do, Dm and Du are bent at right angles, cut short and soldered to pins 14, 13 and 12 respect- ively of ICi. The remainder of the wiring is seen in Fig. 27. Use very thin wire (for instance enamelled copper wire or wire with teflon insulation). The output of FF 2 (pin 1) is connected to where originally pin 5 of ICi was connected. Since pin 5 of the MCI 45027 is an ad- dress input, it is, in principle, possible to control even more than 80 locomotives as originally envisaged. If this pin is con- nected to the logic + line, the appro- priate decoder may be accessed by key- ing in the locomotive address and press- ing the function key. It is possible in this way to control up to 160 locomotives simultaneously. 5.30 elektor india may 1989 CLASS-D AMPLIFIER by J. Bareford The terms digital amplifier, class-D amplifier, switched amplifier and PWM amplifier all refer to a type of amplifier that converts its input signal into a rectangular signal with variable duty factor. The high efficiency achieved by a class-D amplifier makes it of particular interest for mobile and public address applications, where low distortion is not a prime issue. The AF power amplifier described here works from a 6 V batteiy, and delivers up to 5 watts. A such, it is eminent for use in, for instance, a megaphone. elektorindia may 1989 5.31 PWM signal superimposed on the sine- wave is small. Switches as amplifiers The basic operation of the PWM ampli- fier may be illustrated with the aid of the block diagram in Fig. 2. Assuming that the input is short-circuited, switch Sa charges capacitor Ci with a current /:, until a voltage is reached that cor- responds to the upper switching threshold of the electronic switch. This then connects /?? to ground. Next, Ci is discharged to the lower switching threshold of Sa. The resulting square wave has a frequency of about 50 kHz, as determined by Ci and Ri. An AF signal applied to the input of the A well-known prob- lem with mobile AF amplifiers is that their low efficiency makes it virtually impossible to generate high power levels from a low sup- ply voltage. The am- plifier described here has a total efficiency of almost 100% at a distortion level that is tolerable with mega- phones and similar P.A. equipment. The basic principle behind the design is amplifier effectively causes the additional current 1\ to propor- tionally reduc.e or in- crease the charge time, and increase or reduce the discharge time. The input signal thus controls ( modulates ) the duty factor of the rectangular signal which appears at the loudspeaker output. Two further principles are important for the basic operation of the PWM amplifier. First, switch St> is controlled in anti-phase with Sa, and keeps the other loudspeaker terminal at a voltage comple- mentary to that of the PWM signal. This ar- rangement results in a switching power out- put stage of the bridge type: the loudspeaker is driven with the full supply voltage at each polarity, so that the highest possible current consumption is achieved. The second additional point to note con- cerns inductors L i and L These in- tegrate the rectangular signal and so make it sinusoidal as seen in Fig. 4. The inductors also serve to suppress harmon- ics of the 50 kHz rectangular signal. High sound levels from a small circuit The components shown in the block diagram are easily recognized in the cir- cuit diagram of Fig. 3. The input section of the PWM amplifier is formed by a ca- pacitor (or electrostatic) microphone, biased via Ri, coupling capacitors Ci and C*i, a volume control, Pi, and an Pulse-width modulation Figure 1 shows the principle of pulse- width modulation (PWM): the input sig- nal controls the duty factor of a rectangular signal of a much higher frequency. The on-time of the pulse is proportional to the in- stantaneous amplitude of the input signal. The sum of the on-time and the off-time — and, therefore the frequency — is, however, constant. Hence, a sym- metrical rectangular signal (square wave) is generated in the absence of an input signal. In order to obtain reasonable sound quality, the frequency of the rectangular wave must be at least twice as high as the highest frequency in the input signal. A simple low-pass filter may then be used for integrating the rectangular signal. The result is a signal that may be used for driving a loudspeaker. The signal conversion is apparent from the lower oscilloscope trace in Fig. 4. The upper trace displays the output signal after filtering, measured across the loud- speaker. The amplitude of the residual Fig. 2. Block diagram of the class-D amplifier. Fig. 1. Conversion of a sine-wave into a pulse-width modulated (PWM) signal. amplifier based around opamp Ai. The previously discussed switches Sa and Sb are formed by electronic switches ESi to ES-t in combination with transistor pairs Ti-Tt and T2-T4. The part indications for the components that form the PWM generator correspond to those discussed with the block diagram. The unusually high efficiency of the PWM amplifier is perhaps best il- lustrated by the fact that the output tran- sistors remain cool under all drive con- ditions — dissipation in the power out- put stage is virtually nought. When selecting practical inductors for Li and Li, remember that their ability to pass 3 A without becoming saturated is far more important than the actual in- ductance. The inductors used in the prototype were toroid types salvaged from a lamp dimmer. Diodes Dr to D6 limit the reverse e.m.f. generated by the inductors to a safe Fig. 3. Circuit diagram of the 4 W elass-D amplifier for public address applications. 5.32 elektor india may 1989 Fig. 4. Sinusoidal oulpul signal (upper mice) and PWM control signal (lower (race). value. Components Di, C 3 , D: and Rj provide the non-inverting input of opamp Ai with a well-filtered potential equal (o half the supply voltage. As with a conventional opamp-based amplifier, the voltage gain is set by a negative feed- back network. In practice, Ri and Rs set the gain to 83 to ensure adequate micro- phone sensitivity. When high-impedance signal sources are used, Ri may be in- creased accordingly. Because of the phase shift introduced by Li and L 2 , feedback is realized with the aid of the rectangular signal at the col- lector of Ti, rather than with the sinusoidal loudspeaker signal. The opamp itself, in combination with Cs, provides the required integration of the PWM feedback signal. It should be noted that the feedback system reduces the amplifier’s distortion, but not, un- fortunately, to a level that would make it suitable for applications other than public address. A class-D amplifier with low distortion would require a much higher supply voltage than used here, and would be a fairly complex design. This, in turn, would almost inevitably result in much reduced overall efficiency. The electronic switches in the amplifier must be HCMOS types — a standard CMOS Type 4066 is so slow as to cause a short-circuit across T1-T3 and T2-T4, with the obvious risk of overloading or even destroying the amplifier. Bullhorn The class-D amplifier is preferably used for driving horn-type loudspeakers, since these offer the highest sound pressure for a given power level. The prototype of the amplifier was used in combination with a 6 V battery pack and a pressure chamber loudspeaker. The available 4 watts of output power resulted in a megaphone with an im- pressive acoustic range. Four series-connected 1 .5 V dry batteries (HP11; C; UM2; Baby) or alkaline monocells provide the supply voltage for the megaphone. When this is used fre- quently, a rechargeable NiCd or gel-type (Dry fit) battery may be preferred. The maximum current consumption of the megaphone is about 0.7 A, so that an alkaline battery has sufficient capacity for 24 hours operation at full output power. For non-continuous operation, however, a set of dry batteries is per- fectly adequate. Whatever power source is used, the supply voltage for the amplifier should not exceed 7 V, because the HCMOS switches in ICi do not operate correctly any more at this level. Fortunately, the absolute maximum supply level for the amplifier is higher at 11 V. H Parts list Resistors (± 5%): Ri;R4=2K2 R2;R3;R7 = 100K R5=180K Re = 22K Ra;R9= 1K0 Rio. . . R 13 incl. = 68R R 14 . . .R 17 incl. = 39R Pi = 10K logarithmic potentiometer Capacitors: Ci;Cj;C4 = 10n C3=47p; 16 V C5=180p Ca= IpO; 16 V C7 = 330p Ca=4700p; 16 V Semiconductors: Dl;D2= 1N4001 D3. . .Da incl. = 1144148 Ti;T 2 = 80131 or BD226 T3;T4 = BD132 or BD227 ICi =CA3130 IC2 = 74HC4066 Miscellaneous: Si = push-to-talk switch. Li;L 2 = 40pH; 3 A toroid suppressor choke. LSi= 4. . ,8R; 10 W: waterproof horn loud- speaker. Capacitor microphone. PCB Type 87676 f " Fig. 5. Printed-circuit board for the amplifier. elektor india may . 1989 5.33 PRACTICAL FILTER DESIGN (4) by H. Baggott The previous part in this series discussed the most important low-pass filters. This month we turn our attention to high-pass and band-pass networks. Since these are derived from low-pass sections, we often speak of derived filters. A high-pass section is derived simply from a low-pass section by substituting i/jai in the transfer function for jco. This is not nearly as complicated as it looks, and is fairly easily realized in practical terms as well. In practice, it means that in a passive filter all inductances are re- placed by capacitances and all capacitances by inductances. In an active filter, all resistances and capacitances are interchanged. The computation of the new compo- nents is also simplicity itself. First, calculate the normalized values of all components for the low-pass section, re- place the components by their “op- posites” and compute the values of the newly required components as follows. Passive filters: CHP=l/Ll.P Lhp = 1/Clp Active filters: Chp = 1/Rlp R HP= 1/CT.P Once the normalized high-pass filter has been computed in this way, the actual component values are dimensioned for the required cut-off frequencies. Figures 18, 19, 20, 21, 22 and 23 show the high-pass filters derived from the low-pass sections discussed in Part 3. These are: • passive type with equal input and output impedances; • passive type connected to source of negligible internal resistance; • two-pole active type with voltage follower; • a filter with a real pole; • a two-pole filter with variable gain; • a state-variable filter. The interchanging of resistances and capacitances does not work in a state- variable filter. This type requires the ad- dition of a summing amplifier that com- bines the input signal, the band-pass signal and the low-pass signal into a 5.34 elektor india may 1989 Fig. 18. Passive high-pass filters with equal input and output impedances (Ri = Ri.): (a) nn-type; (b) T-type. Fig. 19. Passive high-pass filter connected to a source of negligible resistance and ter- minated in Ri.. high-pass function. In the case of an odd-order filter, the amplifier is followed by a passive RC filter. The computation of the various compo- nents in a state-variable high-pass filter is carried out as follows. First, calculate the normalized high-pass pair of poles: «' = «/(a J +/F) [25] [26] The component values are then arrived at: R\=Rz = l/(4na'C) [27] R ) =f?4 = 1 /2 ti C]/[( a') ! + (fi'Y] [28] Rs=2a'R/(v'[(a') 2 +(iS') 2 ]} [29] Re-AR [30] Fig. 20. Active high-pass sections with opamp connected as voltage follower: (a) two-pole type and (b) three-pole type. where A is the amplification. Wide band-pass filters In the computation of band-pass filters use may be made of a low-pass and a high-pass section that, dependent on the required characteristics — band-pass or band-stop — are connected in series or parallel respectively. This method can, however, only be used where the pass band or the stop band is wider than about an octave. A schematic representation of a band- pass filter is given in Fig. 24. Here, a low-pass section and a high-pass section are connected in series, which results in only the common band (fi — f 2 ) appear- ing at the output. The order in which the two are connected is not important as long as the low-pass cut-off frequency, f 2 , is higher than that of the high-pass Fig. 22. A two-poie active high-pass section with variable gain. section, fi. A schematic representation of a band- slop filter is shown in Fig. 25. Here, a low-pass section and a high-pass section are connected in parallel to prevent the band fi — f: appearing at the output. The major difference between the filter in Fig. 24 and that in Fig. 25 is the fact that in the former the low-pass cut-off frequency is higher than the high-pass cut-off frequency, while in the latter it is the other way around. The sections are computed in the usual way as discussed, after which they are combined. Both the band-pass and the band-stop filter may be passive or active. In a passive type the input and output impedances must be equal, otherwise there will be a mismatch that will ad- versely affect the filter characteristic. In the case of an active filter, the two sections are connected in cascade to form a band-pass filter, and in parallel with the aid of an additional summing amplifier to form a band-stop filter. Fig. 24. A wide band-pass filter is obtained by connecting a low-pass section and a high- pass section in series. Fig. 25. A wide band-stop filter is obtained by connecting a low-pass section and a high-' pass section in parallel. ark Next month’s instalment will deal with narrow band-pass filters. elektor India may 1989 5.35 DESIGN IDEAS NEW CIRCUIT PROTECTION DEVICES FOR LOUDSPEAKER SYSTEMS by Derek Overton* A novel type of circuit-protection device is now available to protect loudspeaker systems from damage. Loudspeakers are generally designed and sold separately from amplifiers. Thus, mismatches may occur that can lead to damage. It should be noted that modern digital recordings often place additional burdens on an audio system. Speaker damage may result from a num- ber of factors, including the following. • High-power amplifiers may simply overdrive the speech coils with ex- cessive power on sustained programme material. • Digital recordings, with their ability to reproduce high-frequency sound, place an extra strain on tweeters. • Low-power amplifiers may be oper- ated in clipping mode, which causes an upward frequency shift of power into the tweeter, resulting in an overload. This problem is especially common with the wide dynamic range of digitally pro- grammed material. • Unstable amplifier systems may oscillate in the ultrasonic range, which overloads the tweeter. Fig. 1. A selection of Bourns MultiFuse™ resettable PTC circuit-prolection devices. * The author is with Bourns Electronics Limited 5.36 elektor India may 1989 Fig. 2. A PTC device in a typical circuit. A new range of Positive-Temperature- Coefficient (PTC) circuit-proctection devices is now available to overcome these problems. The new devices, known as MultiFuse™, act like fuses under over- current conditions, but “reset” themselves by returning to their low- resistance state once they cool below their “trip” temperature. Because of this, they overcome the drawbacks of other overcurrent protec- tion products such as fuse links, bimetallic circuit breakers or ceramic temperature-dependent resistors. Fuse links are not resettable; bimetallic circuit breakers are prone to vibration, welding, sparking, contact-resistance variation and recycling problems; and ceramic devices are slow in operation and may suffer from low-resistance or short- circuit problems under certain con- ditions. Under normal conditions, the resistance of a MultiFuse™ device is comparable to that of a fuse link — between milliohms and a few ohms — depending on the specified current-carrying capaci- ty. When an overcurrent heats it up to its trip temperature, its resistance increases by many orders of magnitude, limiting the current from the power source and thereby protecting the circuit. Once tripping has occurred, the residual current maintains the device above its trip temperature, and latches it in its pro- tective high-resistance state. The device will return to its low-resistance state and reset once it cools below its trip tempera- ture, which is achieved by switching the power off or substantially reducing the current. Once the fault condition has been cleared and the device reset, normal circuit operation resumes. With a MultiFuse™ circuit protection device in series with a loudspeaker (either before or after the cross-over filter), a sustained overload causes the device to switch to a high-resistance state to protect the loudspeaker. A reduction in drive power, resulting either from a change in the music or by the user turn- ing down the power, allows the device to reset automatically. A properly sized device can be put in series with the loudspeaker to be pro- tected as shown in Fig. 2. The device has a low resistance (typically 0.030-0.2 Q for common loudspeaker sizes) and es- sentially no impedance. Thus, the only effect on sound is a slight reduction in drive power (typically less than 0.1 dB). No measurable distortion has been found with normal signal levels. A sustained overdrive condition causes the device to switch to high resistance. The speed of tripping depends on the amount of overcurrent. The resistance level in the tripped state (Rps) depends on the power dissipation of the device (Pd, which is essentially a constant) and the square of the drive voltage, V, specifically: r ps «v 2 /Pd Therefore, once a device trips, an in- crease in drive voltage raises, and a decrease lowers, the resistance of the device. The increased resistance is typically thirty to forty times higher than the base resistance, and this causes an abrupt reduction in power to the loudspeaker when the device trips. The effect on load power is illustrated in Fig. 3. Initially, load power increases with drive voltage. When the drive voltage causes an excess current to trip the device, load power is reducted typically by 20-30 dB. Further increases in drive voltage reduce the load power even more. A reduction in drive voltage increases the power to the load. Since the device is now the dominant load, it does not trace back the original curve. Rather, the drive voltage must be reduced until the device can no longer draw the power (typically 2-3 W) to maintain itself in the tripped state. The approximate condition for the device to untrip is: V 2 /4RL