THE INTERNATIONAL ELECTRONICS MAGAZINE FIECTRONICS S pu BUMPER CHRISTMAS ISSUE With over 50 construction projects Fuhetion® PPR Solid-state FIESTA lam Ne L te _ A model analysis 2.5 ounbe ; yom(xem T CONTEN Ss Solna is egies (among others): + Westart anew series on + PAL test signal generator In nest month's issue We thank all our readers wherever they are for their continued support and extend to all of you our best wishes for ‘understanding electronics PEACE, HOPE, BROTHERHOOD, LOVE AND FAITH “Figuring it out in the New Year ‘+ Mulki-core cable tester + Dil video ampitier! Dal rs = "Covina | #monuer ovsaviay so nies unio «une mee ‘Compressor/imiter 30 + .2GHemotifunction | 32 Woutpur amphifer 2 frequency meter Pa? | Pulse generator for AV recorder oe ~ ‘Low-noise amplifier oid Decibel stepper 82 ‘Speech/sound memory 60 60 W music amplifier 62 Low-noise amplifier II 76 ‘Band-pass filter with extra feedback 8 ‘Copying with A-V recorders 80 Sa ‘Ootput amplitier for ribbon loudspeakers (2) 98 tenes contronets frequency meter/pulse BRED DO Peace eee ie generator. The mstrument pe dale Bh offers microprocessor Riad SER conan ue os 4 control pulse and period measurement. battery option, pulse generator, LD. eal-out and menu-driven COMPUTERS & MICROPROCESSORS Diskette-side chooser ‘operation. ‘The compact, esp: neo portable instrument is a '80C552 microprocessor system Compact RS232 isolator Bees successful frequency meter DESIGN IDEAS we published in 1985.thoo- | Pat ackjack sands of which are in use all ‘over the world, GENERAL INTEREST Seven-eleven Digital audio-visual system (2) digit counter module Pulse-operated relay Single-IC timer PWM governor Real-time switch debounce Digital tape counter BIRELS Se ‘Copyright © 1952 elekiwur BY fi POWER SUPPLIES & BATTERY CHARGERS Unblocking the pump, Voltage inverter oe December 1992 Number 206 Improved LM317 regulator 6 ‘Charging temperature monitor o Mains power-on delay ™ Solar-cell supply n Switched S-V supply % Symmetrical power supply 7 ‘Automatic NiCd battery charger 8 Delta-peak NiCd battery charget 86 RADIO, TELEVISION AND COMMUNICATIONS. Mint crystal oscillator 30 Video demodulator 31 Solid-state T/R transverter interface 32 Pulse generator for A-V recorders ry PLL synthesizer for TV receivers 66 ‘Dummy toads 83 SCIENCE & TECHNOLOGY ‘A model analysis, 9s ‘TEST & MEASUREMENT 1.2 GHz multifunction frequency meter 11 ‘Thermocouple to DMM interface 38 Relative humidity sensor 40 Digital paitern generstor 6s Inductive proximity switch 7 Serial data generator 68 “Temperature to frequency converter 9 Capacitance meter ™ Discrete frequency to voltage converter. 73 MIDI (cable) tester ” Repeating pulse generator 18 Current probe for printed circuit boards 81 MISCELLANEOUS INFORMATION Electronics scene s Events 7 Corrections 35 Readers’ leters 104 New books 107 Readers” services 110 ‘Terms of business m2 Buyers’ guide us Index of advertisers na ELERTOR ELECTRONICS DECEMBER 1992ELECTRONICS SCENE SCOTTISH UNIVERSITY TO EXTEND LABORATORIES INTO ‘THE NORTH SEA ambitious research project, designated Forthbase is being launched by the Department of Electrical and Electronic Engineering at Heriot-Watt University. Edinburgh. ‘For some 20 years, the department has con: ducted substantial research into underwater communications, remote operated vehicle (nov) systems, advanced underwater roboti systems and sonar imaging. Now. its mult disciplinary marine technology research will be extended along the coastline and into the waters of the Firth of Forth estuary. ‘The Firth is representative of many major {estuaries around the world, and as such is an excellent location foranestuary reserach pro: ‘gramme. Its coastline embraces nuclear and ‘coal-fired power stations, cheemical pro: ‘cessing plants dacks, bridges, shippinglanes. oil terminals fishing harbours, cities and towns, There is also an abundance of natu- ral life ‘Avhub forthe project will be provided in ‘new cusiom-buill premises on the Riccarton ‘campus in the form of a remote sensing lab- ‘oratory (RSL). This Will house a compreher fencanl' fo fe sive real-time data-gathering system using UNF, microwave, millimetre wave and laser ‘communication links to various experimen tal sites around the estuary. It will be possi: ble to carry out detailed monitoring and re: search over a wide range of activities, in cluding underwater acousticcommunications, navigation studies, water channel character ization, and water quality and other envi- ronmental studies. ‘The setting upof the rst provides unique opportunity to develop a long-standing re- quirement to set up a fully instrumented set of test ranges in the Firth of Forth controlled from the laboratory. Stretches of water have already been identified with suitable end stations offering ideal ranges for sonar, nay~ igation and communications studies in the presence of noise and multipath interfe; ence. These may be accurately modelled through the detailed knowledge of the pre- vailing water column and seabed conditions. ‘The data-gathering system, referred to as Flashback, uses signals being transmittedand ‘measurements made over signilicant under- water path lengths, at depths that would en= gender realistic problems of multipath inter- Terence over typical seabeds ‘Coded signals originating fromthe rst.will FORTHBASE TEST RANGE ¢ relayed via an electromagnetic wave link to the acoustic sending stations. Signals re- ceived at the far end of the link will then be transmitted back via a similar free-space link in either raw analogue or processed dig- ital form, to be decoded and analysed at source in real time. In addition to the through-water commu nications data, relevant oceanographic and environmental data will beavailable from the remote station for incorporation in the re~ tum signals to the laboratory. Heriot-Watt University, Department of Electrical and Electronic Engineering, Ric- carton, Edinburgh EH14 4AS; Telephone 031 449 S11 NEURAL NETWORK FULLY SUCCESS- FUL IN MATCHING TESPAR DATA combination of a neural network from Neural Computer Sciences and signal processing has achieved automatic real-time pattem recognition. One of the first applica~ tionsof the new PC-based neural network pack- age, called NeuralDesk, is likel the widespread availability ofintelligent, fully automated, signal recogaition systems. The application is by Domain Dynamies, who have RLEKTOR ELECTRONICS DECEMBER 1992used the package to automate the recogni: tion of data from the oulput of its innovative signal processing technique, TespaR—time- encoded signal processing and recognition, Currently available in the form of two cir cuit boards, TESPAR is capable of being con- verted to a single piece of silicon, opening up 2 host of embedded and real-time appli cations such as biometric yerification of in dividual speakersor signaturesforsecrty pur salth monitoring, poses, or machine hi TesPaR provides a highly effi of capturing and storing a single elemental signature” of avoustic activity, which then provides a reference for recognizing pat tems, The technology solves long-standing ni means problenis in pattern recognition. TesPAR isaiprocessor-hased circuit that. itizes and converts sample data, For instance, ing, into special codes that isties of the waveform. The ccode system creates a new digital language for describing and comprehending acoustic information. Itoutputs these numerical codes as matrices. From these samples, a single a person sp denote charact statistically relevant reference, orarchetype, matrix is generated, regardless of the length of data analysed. Up (0 now, comparing this reference 10 new data in order to recognize pattemshasinvolved the use of statistical cor relation techniques. However, because the reference data provided by TESPAR is ex tremely compuct, typically just a couple of hhundred bytes, itis ideal for use in. embed= ded and real-time systems, The neural tiework provides a means of ‘automatically comparing new data with the reference matrix and quickly distinguishing ‘between an input which, although slightly dit ferent trom the reference, is an acceptable match, and data that is definitely at vari ance. The fixed:-size tabuar nature of TeSPAR produced data is perfect tor the input mech: anisms offered hy NeuralDesk, Inall of ist: als, on a number of real Domain Dynamics projects, the neural network has proved 100% makit fact, analysis of the results has shown th successiul i sorrect distinctions. In theseparation of correct and incorrectmatches. isso wide thatthe neural network-rESPAR com: bination isthe ideal technology for perform. ing the task, Moreover, because the circuitry required to embed a neural network is quite modest, Domain Dynamics expected (0 be able to produce a single-chip neural network plus TESPAR solution within | to 2 years. ‘Neural Computer Sciences, Unit’, Lulworth Business Centre, Nutwood Way. Totton, ‘Southampton SO4 3WW; Telephone (0703) (667 775: Fax (0703) 663 730, Domain DynamicsLid, Royal Albert House, Sheet Street, Windsor SL4 IBE: Telephone (0753) 831 424. NEW ERA IN OPTICAL COMMUNICATIONS Trmnarics binned in he Sepember issue of IEE Review, the monthly tech- nological magazine for members of the Institution of Electrical Engineers. Profes- sor Nick Doran of Aston University. describes how engineers and scientists ean now pro duce special light pulses, called optical sol tons, that can travel for hundreds of thou sands of kilometres without changing their shape. The development promises a new cera in optical communications. Over the past 20 years, optical fibre has rev. olutionized the world’s telephone system. By transforming speech into pulses of light. and sending these pulses along ultra-clear glass ‘ore, communicationsengineersean pack hou sands of telephone conversations into a sin Je fibre. But there's « problem: setting more Conversations into the libre means inereasing the number of pulses per seeond which, inturn, ‘means reducing the gap besween successive pulses. Unfortunately, conventional lend to spread aut as they travel alo fibre, limiting the rainimum gap possi tween pulses and the number of conversa tions a fubre can arr The optical soliton is the solution to the problem of spreading light pulses. Because solitons keep their shape, they can be packed closely together, allowing # dramatic increase in the capacity of optical fibre, transforming the existing limit of around ten million pulses per second to several hundred million pulses per second. (Professor Nick Doran is the UK"s leading authority on optical solitons. He is professor ‘of optoelectronics in the Department of Elec: tronic Engineering atthe University of Aston, and previously headed aresearch groupal BT's Martlesham Laboratories) ‘The Institution of Electrical Engineers, Savoy Place, London WC2R OBL. CIRKIT GAINS BSS750 APPROVAL. rkithave been awarded approvalto BSS750, Part 2, with the additional stockists’ sup- plement, which is equivallent to the Euro- pean EN 29002 and International SO9002 Tifleales. as part of the company’s continuous ‘corumitinent to quality and service. The cov= eted award is granted only afier a thorough ‘examinationof the company’ systemssand pro- ‘cedures, and refleets the commitment of the entire workforce to a consistently high level ‘of service, Seen in the photograph is Cickit's quality manager, Andrea Chapman, respon sible for the granting of the approval (Circkit Distribution Ltd Park Lane, Broxbourne EN107NQ, Telephone (0992) 441 306. Utis a real pleasure t he able to show one of the, as yet, far 100 few successful women in the electronics industry ~ may Ms: C apna he.an example ra many others. dor) BSS750 FOR MULTICORE, TOO Nhe British Standart Insitutiouhuas awarded. Multicore Solders Lid.seertfieate ot re istration to BSS750, Past 1. following tailed evaluation of eompany-wide quality sys tems and procedures. Multicore’s chairman and managing di rector said that he believed Multicore were the first manufacturer of soldering materials to have received accreditation to BSS750. Pan, which isalsoinaccordance with EN29001 sand the better known ISO9001 inte standard. Multicore Solders Lud. Kelsey House Wood Lane End, Hemel Hempstead HP24RQ; Telephone (0442) 233 233, ional NEW ‘MOUNTBATTEN MEDAL’ FOR ENGINEERING ACHIEVEMENT ust after this issue has gone to press, The RU Hon Sir Leon Briltan, Vice President of the European Communities, will, on 12 November, deliver the 1992 National Elec tronics Council Mountbatten Memorial Lectureat the Institution of Electrical Engi neers entitled “High Tech in Europe ~ Whiat price a policy?”. The lecture will take in the confliet between EC and national policies, competition issues, industrial policy and the global context, as well as asking who really decides where we go trom here! govern: ment, industry or the consumer? How far hhas Europe got in achieving an internal mar ket? Leitimportanttothink in termsof Europe vs US vs Japan? How can confidence be put back into European industry? ‘After the lecture, His Royal Highness The Dake of Kent, Chairman of The National Electronics Council, will present the new The Mounthatten Medal. This medal is awarded for an outstandi tocleetronies orinformation technology that ‘confers great benefitto the United Kingdom, The Institution of Electrical Engineers, Savoy Place, London WC2R OBL. BLERTOR ELECTRONICS DECEMMER 1992SEVEN-ELEVEN Seven-Eleven is an electronic slot machine. The object of the game is to ‘break the bank’ by accumulating 99+ points. The player starts with 20 points and wins points based on the outcome of the spin of the ‘wheel’, attempting to spin 7's and 1’s. By Larry L. Cameron EVEN-ELEVEN is a simple electronic Jgame project based on the 8748/8749 B74’) series of microcontrollers manu= wed by Intel. Features of the project ide simple construction, entertaining play, and educational instruction for those interested in seeing the use of the 874% in- struction set in an application. This article will detail theory of operation and use of the 7-11 ‘one-armed bandit Returning to the actual game, each spin fn the 7-11 machine costs 1 point and the player is subsequently awarded points ‘based on the following table: Spin Payoff Prob. Note 71 inorder 04% jackpot! For exam a spin of 631 would pay off wnd a spin of 113 would pay off cone point, 10 points, The ‘probability’ column indi- cates the percentage chance that a particu- lar spin combination will be seen. The probability of getting any winning spin is 47.1%, so just as with a real slot machine the odds are against you! The game ends when your score is zero (after a spin with- out a pay off) Theory of operation The power of the microcontroller lies in its ability to easily coordinate hardware to software with a minimum of support com- ponents. In this project, an 8748 controls the operation of the entire game, si debouncing the switch signal, keepin track of the player's score, and controlling the spinning wheel displays with a mini mum of support circuitry. Two 74LS48's Fig, 1. Seven-eleven is an electronic game based on an 8748 microcontroller ELEKTOR ELECTRONICS DECEMBER 1992SEVEN-ELEVEN are required to display the score, but all other aspects of the game are controlled. exclusively by the 8748. The following dis- cussion assumes familiarity with the pinout and operation of the 874x micro- controllers; please consult an Intel data: book for more information if you need to When power is first applied, the game scrolls the numbers 711’ on to the wheel display, initialize variables, and then en- fers a waiting period, decreasing a register value and waiting for a high signal on processor line T0. The register is used to seed a random number. When a high is seen on TO from the switch, $2, the soft- Ware waits fora few milliseconds and then checks again to makes sure the signal is still high. This serves to debounce the high signal. If the signal is still high, the soft- ‘ware decrements the player’s score (wo BCD numbers), and latches these numbers on its PLO-PI.7 outputs which are de- ‘coded by the 74LS49's, and displayed, The number wheels are then set in motion until all displays have stopped spinning, ‘Next, the software checks to see ifthe spin has a payoff, and if not enters the waiting. period again to seed a new random num ber, and repeats the above. If the player's score is zero, no more spins are possible, and the software goes imo 9 eontinual loop, horizontally scrolling the numbers ‘711’ on and off the ‘wheel displays. The game is lost at this point, Ifthe spin has a pay off, the winning digits blink three times, and the player's score is increased by the appropriate amount as per the pay off table. If the player's scare reaches over 99- points, the software also goes into a con- tinual loop with the scrolling '711" display, and the game is won. If the score 1 not ‘over 99, the software repeats the above by waiting fora random seed, ete The three wheel displays are multi- plexed to the microcontroller. To minimize hardware, the controller's BUS latch out- puts are used, along with the P20-P23 toutputs to switch the appropriate display off ane on wa three n-p-n transistors oper- aling in @ saturated mode. For example, to display ‘711’, the appropriate bus lines that correlate to lighting a. "seven are dr ven high, P20 is driven high to forward bias the transistor, and the segments for the first wheel display are lit (the common sathode of the display 1s grounded). Line P20 goes low to turn off the transistor, and the display extinguishes. The bus lines change state to reflect the next number, ‘one’, and P2.1 goes high. The second dis- play lights up to display ‘one’. The transis tor is turned off, and the same process repeats for the thitd displayed number. By repeating this process at a fast rate, the i+ lusion of a constant (‘static’) display is achieved. Multiplexing the displays in this, ‘way also eliminates the need for any cur- rent limiting resistors, For sound, a buzzer is connected to the dot point display, and ADDR VI.T Hex 09 (IsTINe on00 0010 0020 0030 ‘0040 0050 0060 0070 0080 0030 ‘OOAD 0080 ‘ooco 000 O00 ooro 100 Bio 0120 0130 0140 0150 0160 0170 0180 0190 O1AD 0180 o1co 0100 O1e0 O1FO 0200 0210 0220 0230 0240 0250 0260 0270 0280, 0280 0240 0280 02c0 0200 020 o2FO 0500 0310 0320 0330 0340 0350, 0360 0370, 0380, 0380 O5A0 0380 03co 0300, 03E0 O3F0 controlled via P29 and another transistor, Fig. 2. Mieracontroller program in the form of a hexdump. ELEKTOR ELECTRONICS DECEMBER 1992being continually forward bi the wheel spins, and turned off for the resi of the game's functions NERAL INTEREST ed while Construction hints The introductory photograph shows the author's prototype of the game. For those unwilling to program their own 8748 with the aid of the hexdump given in Fig. 2, the author can supply ready-programmed controllers at $12.00 each, Also available at 54.00 isa disk formatted in IBM 360 KByte formal containing both a binary fle and an Intel hex fle of the object code, ready to be loaded into most IBM PC controlled pro~ grommers. If you would like to purchase any of these items from the author, please your cheque payable to Larry and mail to my address. Call 7983 before 5 p.m. central stan and time, for mare information For a professional looking project, i is best to produce a printed circuit board which you may want to design yourselt Advanced constructors may also considet building the project on veroboard or strip board. as the component layout is not par Heularly eritical. Alematively, contact the author for a suggested PCB layout (not given here). Untortunately Fean not pro- vide for sale a pro-ctched, pre-drilled board, but a simple method 1 use fre: quently to make boards is the “TEK-200 film method (also goes by other names), ‘which involves photocopying the PCB lay- ttt to be etched on to a special sheet of mylar film, ironing the subsequent film on toa copper-clad board, and etching Most of the parts for 7-11 can be bought at Radio Shack (Tandy), or at any other well-stocked electronic parts store, IF you can not find an SPDT momentary switch, you may substitute an SPDT toggle switch NEW RANGE OF PROGRAMMER/EMULATORS Technology have announced the launch of a new range of universal pro: gcammers with buill-in emulation cupabil- ities, offering superb capabilities for the design engineer at a very affordable price While other progranimer/emulators ean, only support EPROMSs, the Speedmaster LO00E and Micromaster L000E ean pro- gram EEPROMs, serial EPROMs. NVRAMs, Flash, PALS, GALs. EPLDs, PEELs, Machs, MAPLs. ele...and can em- ulate ROM and RAM up to 128 KBytes {equivalent to 1 Meg EPROM). The pro- grammers are housed in an attractively styled high-densily polyurethane enclo- sire, and come complete with software, manual, printer port cable, power supply adaplor and emulator cable. The printer cable plugs straight into the standard par- allel port of anty IBM compatible PC with: ‘out the need for any expansion card. More dotails ean be abiained from ee 4 necessary. You may have to hunt around for switches that fit in the PCB layout holes, or simply modify the board by drilling holes and adding jumpers to fit your switek. Trecommend the use of sockets For the ICs, and 7-segment displays. 1 used five Radio Shack 7-segment common cathode displays, but any other display could be substituted as long as it conforms to the same pinout as the Radio Shack ones (see circuit diagram). A piece of red cellophane laped over the displays provides for good contrast, and allows you to see the lit seg- ments very well in daylight You may substitute 7448's or 74C4%s, but I found the 74LS48's to be the cheapest variety. If constructing the project on to 3 printed circuit board produced from the author's layout, please note that the 7A1S48's are oppositely aligned versus the pin 1 of the 7ALS48's 3s in the lower left corner of the socket, but pin 1 of the controller is on the upper right side. This was done to facilitate design of a single-sided PCB with a minimun number bf jumpers on the component side, Be reful to correctly orient pin 1's, etc transistor orientation, anodes, cathodes of polarized parts, etc, as you fit compo: The author's game is powered by a 9-V dc, wall adaptor transformer, but it ean be powered from a 9-V PP3 battery. The ctt- ‘cuit draws rather a lot of current (approx 130 mA), se for long, periods of operatic an adaptar is really the only option, Author address information Larry L. Cameron 7020 Grand Canyon #243, ‘Austin, Texas 7% USA. Telephone: (512) 467-9532. NEW PRODUCTS ICE Technology Ltd., Unit 4, Penistone Court, Stat is Yorkshire S30 6HG. Telephone (0226) 7674014. Fax (0226) 370494, " COMPONENTS LIST Rosistors: All resistors are 1 walt, 5% 4 10K Rina 15 9300, AS-IS Capacitors: 2 39pF ceramic C1;02 3 100nF ceramic dise O3:Ca:05 1 TUF 16V electrolvic C8 1 10WF 16V electrolytic C7 ‘Semiconductors: 1 8748H, preprogrammed (S00 text) ier 74LS48 icaica 7805 Ica 2N3904 THT LED 0.3" 7-segment common cathade LED display. Radio Shack 9276-0758, LD1-LD5 aaa Miscellaneous: 1 SPDT switch, onon St 1 SPOT switeh, on-none-(on) momentary s2 7 Wire link 22AWG s7 8.579MHz crystal TAL 5:V buzzer, Radio Shack 273-074 40-pin IC socket (optional) 114-pin IC socket (optional) {16-pin IC sockets (optional) 3-V battery with clip (optional, 8-VDC mains adaptor Printed circuit board (see text) Enclosure Be aa SHORT HEADERS— HIGH PIN COUNT New from 3M Electronic products is the 2500 series of shrouded, low-profile, box headers for bourd-to-board and wire-to- board applications. The 2500 series is available in IT pin counts (10-60 ways), tffering design engineers one of the widest choices of pin count options. The headers are based on 9 0.1 in.» 0.1 in matrix 3M Electronic Products, 3M UK ple, Market place, Bracknell, Berks RG12 1JU, Telephone: (0344) 426726, EKTOR ELECTRONICS DECI HER 19921.2 GHz MULTIFUNCTION FREQUENCY METER PART 1: CIRCUIT DESCRIPTION The instrument described in this three-part article is one you can not afford to give a miss since it is very likely the most advanced frequency meter/pulse generator you can build yourself. Sporting microprocessor control, pulse and period measurement, a battery power option, a built-in pulse generator, LCD readout and fully menu-driven operation, the instrument is compact and portable, and a worthy successor to our famous 1985 microprocessor controlled frequency meter, thousands of which are in use all over the world. Design by B.C. Zschocke HE use of a powerful microprocessor gave an interesting turn to the devel: lopment of the present instrument: what was but a simple frequency meter at one stage evolved into a multifunction test in- strument that can be built at a fraction of the cost of a commercial equivalent. The basic thought was this: if the simple in- strument is capable of measuring. fre quency and pulse duration, why not use the microcontroller’s power to reverse oO /FRONT COVER \ PROJECT) these functions, and output programma ble signals as well? After all, the design was already quartz-controlled, and the ‘measuring algorithms available. Thus, by extending the control software, the origi nal frequency meter was turned into the multifunction instrument described here. The final result is a versatile, accurate and extremely sophisticated piece of test equipment you would, of course, only ex: pect fo see described in Elektor Electronic: ELEKTOR ELECTRONICS DECEMBER 1992 A brief list of thing like this: © Universal counter and signal generator: rowered by mains or by batteries; ‘ompact and portable Iphanumerical two-line LCD readout; @ Menu-driven @ Serial int @ Low cost ce for connection to a PC ‘no expensive components, The choice of the microcontroller to use in the instrament was not @ difficult one Intel's 80C32 is a CMOS controller with very low power consumption, which al lows the instrument to be battery. pow ered, The 80C32 has the ports required to drive a two-line LC display, to interface to the measurement peripheral circuitry, and to convey data via a serial interface. The ports allow the additional hardware to be reduced to a couple of gates, an EPROM from which the control program is run, an analogue signal shaping section, and an LC display. The LCD is used to display data as well as the menu that assists the user in setting up the instrument parame. ters for a particular type of measurement or signal generator modeTEST AND MBASUREMENT CONCISE SPECIFICATION Features: 2 inputs with AG/DG selectable input 1 TTL input GATE output (short-circuit proof) S232 serial interface for PC control Extension port Battery or mains powered ‘Two-line (216 character) alphanumerical LCD readout Single of continuous measurement Compact size (2011 3.5 em) and lightweight Excellent price/pertormance ratio Measurement modes: Frequency meter: 1 mHz to 1.2 GHz Period (1/Irequency): 1 ps to 4.000 Revolution counter: 0.001 to 4x10¥ rev/min Event (pulse) counter: 1 10 approx. 410° Zera counter: 110.410" Generator modes: Timer: 1 1s to 4,000 s Pulse generator’ 8,500 s (0.117 mHz) to 4 us (250 kHz) Duty factor: 1:1 to 1-410" Period number: 1 to 4:10" User-defined settings: Single/continuous measurement Floating gate time Inlermediale measurement result display Period analysis (accurate measurement of low fraquencies at short gate periods) Quiescent level Pulse polarity Sounder oniotf Test routines: LOD test Main test Serial /O test Note: mH2—=smilinertz (0.001 Hz): KHz dlohert2 (1,000 Hz); MHz = megahertz (10° Hz); GHz = gigahertz (10" H2y: ps = picoseconds (10's). bbe measured is lower than 1/24 times the oscillator frequency. Obviously, this will not do for a GHe irequency meter, hence the use of a prescaler to overcome the problem. The prescaler is shown in the lower left-hand comer of the block dia gram in Fig. 1. Two different input signal paths are- possible: either via channel A and a GHz prescaler (IC), or vio channel Band a preamplifier, Ti-Ti-Ts. A multi plexer selects one of the signals, and out puts it to a further prescaler, ICs, Actually, ICs should be labelled ‘pre-counter' rather than "prescaler’, but this may be less tamil: lar. The carry output of ICs ensures suffi ciently long pulses at the microcontroller timer/counter inputs, TO and TI. The op- lied by the microprocessor via port line PILZ. The eration of prescaler ICs is contre prescaler enable signal is alsa the counter gate signal, which is amplified by T~Ts and output to the GATE output of the in strument. This output also supplies gate pulses when the instrument is used as a signal generator ‘Channel C serves to set up different trigger conditions (posilive or pulse edge), and is connected to the inter- rupt inputs of the microcontroller. This al lows the selection between the trigger conditions (positive or negative signal edge! to be made in software The operation of the frequency meter is based on the reciprocal pulse count princi- ple. A gate time is set up during which the Input frequency and a reference frequency are measured, The reference frequency is supplied by the third counter contained tn the 80C32. The input frequency is c lated by dividing the two counter states, and subsequent multiplication oF the re= Provision has been made to test the in: fl strument in steps. Test routines are imple- hor 5 eo mented in the software to help those of you who get stuck with a certain fault dur F On aed? bed ing te contraction, and donot know |S Ae whether the cause is hardware or software pe aS, LATE (this is often a problem with microproces: sor controlled equipment), The test rou tines should help you locate the error in a step-by-step manner, since they addn sub sections ofthe cuit na rater clever way. ore O4 =f 4 Vide the gate timo, the second tn count swofe Loatetso | pukes. Unfortunately, there is a snag. in this set-up; the highest frequency that can rugs [Gp mae Fig. 1. Block diagram of the multitunction {requeney meter. EKTOR ELECTRONICS DECEMBER 199212 Gite Muntuscrion rrrouesey werent HE Fig. 2. The functional blocks shown in Fig. 1 are easily found back in the circult diagram ofthe instrument. Note that the shaded sections are tenclosed in a metal screening to prevent spurious radiation affecting the measurements. cull with the reference frequency. The Circuit in detail Provided the gate time is properly syn chronized to the input signal, this system Since the elements that make up the block allows very low frequencies to be mea- diagram (Fig. 1) are fairly easily found Sured with great accuracy. More about the back in the circuit diagram (Fig, 2),wecan and Dis, limit the input voltage swing 10 frequency measurement principle ina sep- limit ourselves to noting the ‘not So obvi +0.3 V maximum. It is important to note arate article instalment! ous’ only that a number of equivalent or near-equiv ELERTOR ELECTRONICS DECEMBER 1992 OS12 GHz MULTIFUNC ON FREQUENCY METER - 1 COMPONENTS LIST Re 2 RIRS 1 Re 1 RS 2 470 RAAB 1 10k Re 1 2k0 a7 1 56k0 Ra 2 tka RIO.RIG 2 2200 RIt:RI2 + 5600 Rig 4 1500 R14 4 3300 Ris 1 4407 RIT 4 10k presetH PY 1 25kQ presetH —P2 1. 20k0 potentiometer linear, 6mm spindle dia. P3 1 100k presetH PA Capacitors: 1 1uF 16V a + 40pF trimmer G2 1 20pF 3 1 470pF C4 1 WF eV3 radial = C5, 1 220uF 16V ce 1 t00uF 6v3 c7 10 100nF Ca.c13,C18; Cr9.c21 1 39 10V tantalum C14 2 10QuF 25V 615,020 1 47 crs 1, 220uF 6V3 tantalum C17 1 120pF C22 1 1nF C23 1 820pF 24 1 10nF cas Semiconductors: 40 1N4148 D1,02,04;D5; 6,08,010; D14:14:015 1 LED, green, mm dia. 03 1 18V.0.4W zener diode o7 1 1N4001 De 2 BATE? D12.013 1 BF494 a 2 BS170 Tear 3 BS250 T3767 + BF98t 4 1 BF450 5 1 BF324 9 2 74HCTI92 Icrjica + socsz Ice 1 74HCT139 1c3 1 744590 es 1 74HCTS73 1c8 1 270256 EPROM (ESS 6141; see page 110)1C7 Lost Ics SP4731 or SP4633 (Plessey) Alternatives: SDA4212 (Siemens); Us64B (Telefunken); SABG456 (Philips Semiconductors) 1C10 1 7805 Ic9 Miscellaneous: 1 9-V PPS battery Bt 1 passive 8V piezoceramic buzzer B21 1 18-way sub-D socket KT 1 combination of 14-way SlL header and socket K2 4 BNC socket; single-hole ‘mounting Ka-K6 1 16-way boxheader KZ 1 Away Sik pinheader KB 1 way IC pin strip or SIL socket Ko 11 Gay IC pin strip or pin header K10 1. mains adaptor socket. K11 1_ Sway sub-D socket _K12 PCB mount press-key (Muitimec 2CTL2) si Press-key (Digitast) _ S2-S6 Black keycap (w=12mm) Fed keycap (vr=12mm) (ENTER) ‘SPST slide switch S788 S-position side switch $9 42MHz quartz crystal x1 LCD moduie 2x16 characters; 1 row of 14 connections. Preferred type: LTN211F10 (Philips Components). Alternatives: LMOt6L (Hitachiy; EA-D16025AR (Seiko-Epson), ‘ABS enclosure; Bopla EG2030 94V battery holder; Bopla BESO 9. battery clip with wires ‘M2.5x 16mm or M2.5x20mm screw M25 nut M3t6mm or M3:20mm screw M3 nut Plastic PCB spacer, 10mm long IC sockets for 101-IC8 and IC10 30 Solder pins. 1mm dia, + POB plus software, order code 920095 (see page 110) + Front panel fol $20095-F ‘Suggested component suppliers for this project: Cricklewood Electronics (081) 4520161 (passives and semiconductors): Electrovalue Limited (0784) 33603 (pas- sives, semiconductors and Siemens pars); C- Electronics (fax +31 45 281877) (passives, semiconductors: case; switches, LCD and kits). These are not exclusive suppliers: others may also be able o help. alent ICs may be used for the GHz prescaler, Unfortunately, some of these, including the SDA4212, Uo64B, SAB6465 and SPA731, have a tendency to oscillate ‘when no input signal is applied. The in strument then indicates a random fre~ quency. This effect is normal, however, se for alarm, Other ICs, like the SP4633, are stable in the absence of an input signal, however offer insufficient amplification at signal frequencies below 50 MHz. or so. An RF transistor, Ts, con >rescaler output into TTL level Since the GHz prescaler IC has a typical t cd order of provision has been made to switch and no 6 sumption of the it off when it is not used, This is achieved with the aid of FET Ts, which is controlled by the microprocessor via port line Pl. ans that the current reduction is achieved via the instrument's user menu. Input B is intended for input signals with a frequency below about 20 MHz. Input selection switch $9 on channel B al: ELEKTOR ELECTRONICS 0 Jows you to choose between (1) no pream: plifier (TTL input); @) a.c. coupled or @) d.¢, coupled, Depending on the switch set- the channel B input signal arrives ei: ther at pin 13 of ICig, or at gate 1 of DG-MOSFET Ts (via Rio and protection diodes Du-Dit) Potentiometer P3 allows the operating point of the MOSFET ampli- fier to be adjusted depending on the re quired sensitivity. Transistors Ts and Ti give the amplified an limited signal a TTL swing. The amplifier is quite sensitive to supply voltage fluctu J rectangular-wave ations, whence the presence of decoupling 1 Cis. Like the GHz I converter can be switched off to reduce current consump- tion. th FET, Ts, Preset P2 serves to limit the span of the sensitivity potentiometer, P Input C is not followed by a preampli fier. To still ensure the highest possible sensitivity, one input of Schmitt trigger gate ICia is raised to a potential just be capacitors Cir prescaler, the TTL le switching transistor is again a CEMBER 1992 tween the higher and the lower switching. threshold. This potential is adjusted with preset Ps. Note, however, that this ‘raised {quiescent level’ is only effective when Ss is set to ac. coupled. Components Ris and Du-Dis protect the input. The channel A and B input multiplexer is built with a part of demultiplexer IC The channel A input signal, shaped and inverted by ICic, is switched through to one of the multiplexer outputs 0-3 under the control of the signal applied to the en able input (pin 15). The relevant output is selected by the logic level combination ap plied to the 0-1 inputs, Since the input sig nal is connected to the ‘0’ input, the output togeles (pins 12/11). When the enable sig nal is low, output 0 supplies a copy of the input signal, and output 1 the inverted input signal. When the pulled high, all demultiplexer outputs are high. At the same time, channel B is en- abled via ICid. The channel A and ‘channel B signals are combined in ICs, enable pin is|| oz | vor 6 zo ea so }_- A 0, —_____________| (wus svoyouousp ne) a St-1- 60026 Fig. 4. Driling template forthe front panel of the instrument, Schmitttrigger gates ICi and ICs form a monostable that serves to lengthen the carry pulse supplied by external counter ICs. The pulse length is deter mined by Cs and R3. Diode Di and resistor Rz protect the input and output of this cir cuit against the small peak at the end of the pulse, If the carry pulse is longer than the monostable time, the monostable effec tively does nothing, Transistors T? and Ts provide the ne essary drive at the GATE output of the in- strument. Resistors Ri and RS limit the is of the FETs to the extent of making the output short-circuit The TTL-to-RS232 converter is built around ICs and T2. It does not require a drain-source currei separate +12-V supply since it is powered by the PC’s RS232 port. If this is initialized and not in use, the TxD (transmit data) line is at -12 V, and the RTS (request to send) line at +12 V. A supply voltage of -12 V is built up across capacitor C20 (via diode Ds), and a supply voltage of +12 V across capacitor Cis (via diode Ds). Diodes D2 and Ds protect the circuit when the RS2. ELEKTOR ELECTRONICS DECEMBER 1992Fig. 5. Lacking an oscilloscope, build 1 simple adaptor circuit to test a number of sub-seetions of the circuit. interface is not connected, or when the port is incorrectly initialized on the PC. ICs converts the microcontroller’s serial ‘output pulses into an RS232 compatible signal. The voltage drop across LED Ds is about 2 V, which serves as the reference voltage. The TxD signal from the PC is first limited to +15 V (max.) by resistor Rs and diode Dr. Next, itis inverted by FET 1.2 GHz MULTIFUNCTION FREQUENCY METER + 1 Tz, and fed to the microcontroller’s RXD. input. An external pull-up resistor is not required since this is contained in the 30C32. ‘The instrument contains a piezoceramic buzzer, Bz1, of which the drive is slightly unconventional. The buzzer is a passive type connected to address line AIS. This line is not decoded, which means that the content of the 32-KByte large EPROM is ‘mirrored’ in the upper half of the 64 KByte address space. This circumstance is exploited by to enable the microcontroller to drive the buzzer by toggling A15. Construction ‘The instrument is built on a double-sided through-plated printed circuit board of which the artwork is shown in Fig. 3. This PCB is available ready-made, together with the control software in EPROM, through our Readers Services (sev page 110). To prevent noise and spurious radiation from the digital section affecting, the measurements, it is necessary to fit a metal screening around the channel B am- plifier and the channel A prescaler. The relevant sections of the circuit are ink cated by a background shade in the circuit diagram, Returning to the PCB, start by breaking off the keyboard section. The space so cre ated in the enclosure is used {0 fit the 9-V battery holder later. Check that this can be done using the unpopulated board, the case and the battery holder. The connect ing wires of the switches must be cut off as far as possible because the keyboard PCB is fitted on top of the main PCB later. ‘Make sure that the clearances in the en- closure for the keyboard, the LCD and the slide switches, as well as the hole for the spindle of potentiometer Ps (LEVEL BY, are cut and drilled before actually fitting and connecting these parts. The same goes for the small Holes that give access to the LCD contrast preset, Pi, and the reset switch, SI. If you can get hold of low-pro- file parts you may want to fit the preset and the switch at the track side of the board, and make holes in the bottom half of the enclosure. ‘The cutting and drilling details for the plastic case are given in Fig. 4 After finish- ing the mechanical work on the case, take the populated keyboard and the LCD ‘module, and fit them provisionally on the main board using plastic PCB spacers Adjust the length of the spacers until the LCD and the keyboard are at the desired height (try the ‘fi by mounting the top half of the enclosure a couple of times). ‘On the main board, fit all resistors, all diodes (except Ds), and all transistors (ex- cept TH), Take care to avoid short-circuits with the screening around the prescaler and input preamplifier section. Next, con centrate on transistor Ts, which must be fitted as close as possible to the PCB sur- face. Also note the orientation of this MOSFET: depending on the manufacturer of the device, the drain is usually the longest terminal of the four Proceed by fitting the IC sockets, fol- lowed by the remainder of the compo- a Fig. 6. Front panel foll design shown reduced to about 62% of true size. The ready-made three-colour self-adhesive foil is available through ‘our Readers Services. ELERTOR E1 CTRONICS DECEMBER 1992nents. To enable the LCD and the key- board to be fitted properly, it may be nec- essary to cut off the small plastic studs at the underside of trimmer capacitor C2 LED Ds is mounted like a resistor, ic. straight on to the PCB surface. It has no function as a visual indicator, and is ob- scured by the LCD module above it The keyboard does not need connecting wires. A length of wire wrap style IC pin strip may be used to make a o-way pin header (Ks) and a 6-way mating socket, Kio, Note that the pins that form Ko must be mounted at the solder side of the key- board PCB. The display connections are made in a similar way. The LTN211-F10 LCD module (from Philips Components) used in the prototype was fitted with a 14- way SIL socket, of which the connecting, ‘wires were left at a length of about 3 mm, so that the body of the connector is at about 5 mm of the underside of the LCD board. This socket mates with a l4-way pin header, K2, on the main board. Note that such a direct connection may not be possi ble with other LCD modules, in which case abit of wiring may be necessary TEI, Fig. 7. Wiring details for slide switch $9. With the exception of $9, Ss and Ss, all switches are fitted direct on to the PCB. So is a two-pole three-position slide switch of which the wiring is given in Fig. 7. Points B, Cand Dare solder pins at the PCB edge. Point E is a solder pad at the PCB under- side, next to transistor Ti. The same goes for point F (solder pad underneath IC). Be sure to use sereened cable here. At this stage, you are ready to fit the screening around the prescaler/preampli fier input. Bend an approximately 240-mm Jong, 13-mm high, piece of tin plate or brass around the solder pins in the four comers of the PCB section, as indicated by the component overlay. Solder the screen ing to the comer pins. Next, cut and bend a cover for the screening using the same plate material, Drill holes in the cover to ‘enable the spindle of P3 to pass, and to give access to presets Pz and Ps. Do not ‘mount the cover on to the screening until the circuit has been adjusted and tested ‘The two slide switches, Ss and Se, may be glued on toa small bracket which is sol- dored to the screening, Most of the above construction details are illustrated in the photograph of the opened prototype. First test Give your construction a thorough inspec- tion Before connecting the supply voltage. At the component side, look for incor rectly oriented polarized parts (diodes, electrolytic capacitors, transistors, ICs), At the solder side, look for dry joints, splashes and hair joints between solder spots. ‘Tum presets Pi and Ps fully clockwise, and P2 fully anti-clockwise. Turn the po tentiometer, Ps, fully clockwise. Bit the keyboard PCB and the LCD module, connect the power supply (a ‘mains adaptor), and switch the instrument fon. Adjust the LCD contrast preset, Ps, ‘until the text GHertz counter START or MENU can be read on the display, What to do if these two lines do not appear? Not to ‘worry, the test routines built into the in- strument will come to your aid. All you need in addition to the following text is a multimeter and an oscilloscope. If you do not have an oscilloscope, build the small test adaptor shown in Fig. 6. A PC is re- quired to test the RS232 interface of the in- strument. Test routines The frequency meter must be switched off immediately if the power-on message does not appear on the display. Disconnect the power supply and check its output volt- age. If this is okay, connect it to the instru- ment again, switch on, and measure the current consumption. Any value greater than about 100 mA indicates a fault, the ‘most likely of which is that the supply voltage is reversed (when Dy conducts). If this is not the case, the fault may be a short-circuited copper track, or a defective component For the following tests it is assumed that the CPU, the EPROM, the quartz crys- tal and the address latch function cor- rectly. Display indications are printed in italics. The test routines can be divided into three main groups: LCD test, main board test, and serial I/O test. The instrument is switched to the test mode by keeping the MENU or ENTER key pressed at power- fon, whereupon you enter a wait routine (End Test with >> BREAK <<). From this wait loop, you can call up a group of test routines by pressing the appropriate key (see below). The wait loop is left when the ENTER and MENU keys are pressed si- multaneously (BREAK), whereupon the instrument switches to its normal mode of operation. Within the groups, the individ- ual tests are completed one after another. You can move through the tests by press- ing the ENTER key. Flowever, the Key-test ‘can be left only by a BREAK, ie., by press ing MENU and ENTER simultaneously. LCD test (Group 1) Pressing the START key takes you into the LCD test. Alternatively, you can enter this test directly by keeping the START key pressed at power-on. Provided the LCD works (adjust the contrast with Pi!), the text LCD-test: press ENTER to go on ap- pears. After pressing ENTER two times, the instrument enters the test for ICs. This test causes outputs (and 1 of [Cs to pro duce tones that can be heard when the test adaptor is connected. These tones are visi- ble on an oscillascape as rectangular sig- nals with near-TIL swing. The next test routine drives the inputs of Ix alter- nately. The resulting tone at the output of the gate is two times as high as the input LEKTOR ELECTRONICS DECEMBER 1992tones, In this test mode, the LCD indicates ‘an irregular pattern The last test in the LCD group feeds all characters contained in the character set to the display. The characters move from the tend of the lower display line to the start of the top line. Since the user-defined charac- ters in the set have not been set up at this stage, it is possible that random patterns appear between the standard ASCII char- acter set and the Japanese character set. For the following tests, itis assumed that the display functions correctly Main board test (Group 2) ‘The Main test group is entered by pressing the STOP key. Alternatively, you can enter this test directly by keeping the STOP key pressed at power-on, The first test, Koy-test, serves to check the operation of the keyboard section. It [prompts you to press a certain key, and in- dicates its function in the RUN mode. Doa BREAK to leave this test. The next test the Gate-Out test, which should produce fan audible signal on microcontroller port line PL.?. This signal can be traced right up to the GATE output with the aid of the test adaptor or the oscilloscope. IF it arrives at the GATE output, the wiring of PI.7 and the transistor output driver are all right. Next, the Chunnel-A Test switches on the GHz prescaler via To (+5 V at pin 8 of Cio, and at the emitter of T9). The channel switch is set to channel A. At the same time, a frequency of 971 Hz is generated at the GATE output. This signal may be used to test the channel selector and To, This re- quires switching off the instrument, and removing the prescaler, ICiv, Connect the GATE output to pin 7 of the empty IC socket via a 100-nF capacitor, Switch on, step through the test menu, and run the test, Check that an audible tone is present at the following points: pins 8 and 10 of ICic, pins 12 and 14 of IC2b, and pins Vand 3of [Cas During the Chanvel-B Test, the channel selector is switched to channel B, and the associated preamplifier is switched on via Ts G5 V on Ci?) The GATE output sup- plies a 971-Hez test signal, which may be ‘connected directly to the channel B input. ‘The test tone should be audible at transis- tors Ti, Tsand Ts, at pins 13 and 11 of ICis, and at pins 2 and 3 of ICés, For the Channel Test it is necessary to connect the GATE output to the channel C Input, with the input coupling set to DC. If the display shows ?? behind [CTA or ICTB, either the connection between channel C and the GATE output is at fault, or the rel- evant integrated circuit, When everything, is in order, the display shows OK. The Pre-counter Test serves to check the function of ICs, IC and ICte. Since the GATE output is used to control the pre- counter, input B must be supplied with an external frequency. This may be the ALE signal (CPU pin 30, EPROM pin 20, or ICs pin 11). The input frequency divided by ELEKTOR ELECTRON 256 is then available at pin 9 of ICs, and at pins 4,5, 6,8,9 and 10 af ICs ‘The last test in this group is the Busser test, which should cause the buzzer to pro- duce an audible tone when the HOLD key is pressed. Serial 1/0 test (Group 3) This group of test routines is entered by pressing the HOLD key, and serves to check the operation of the RS232 interface, Alternatively, you can enter this test di- rectly by keeping the HOLD key pressed at power-on. To be able to use the tests, connect the instrument to a PC on which a communi- cation program is run. This program should initialize the RTS output. The PC and the counter should not have any other electrical connection than the RS232 link In a follow-up publication we will de scribe an optocoupler circuit for insertion between the PC and the frequency meter. Such a circuit guarantees a potential-free ‘coupling in all cases, The data transfer rate is set to 2400 baud. After the PC has completed the initiization of the RS232 port, you should be able to measure a positive voltage across Cs, and a negative voltage across Cau, These voltages are measured with re. spect to ground, and must be consiclerably greater than 5 V. Pin 3 of ICa should be at about 2 V with respect to ground. During the SIO Send Test, the instru ment sends a continuous string of BCZ characters to the PC. In case the characters are transmitted too fast, you can slow them down by pressing the HOLD key ‘The SIO Send routine checks the function ‘of opamp ICs, ‘The SIO Receive Test is used to verify the operation of the RS232 receiver cir- cuitry’ in the instrument, in particular, T2 Characters you type on the PC should be displayed on the LCD. ‘That concludes the descriptions of the test routines built into the frequency meter, The test mode may be left by doing a BREAK Alignment “The alignment of the frequency meter is very simple and should not present prob- lems, First, set all presets, arid. the poten- tiometer, to the positions mentioned in the section ‘Fist test’. Adjust Pi until the LCD contrast is optimum, Next, turn Ps fully clockwise (full amplification on channel B). Measure the collector voltage of Ti, and adjust P2 for a reading of 2.3 V. Next, carefully turn Ps counter-clockwise tantl pin 3 of ICta changes from high to low. Measure and record the voltage at the wiper of Pt with the aid of a high-imped- ace voltmeter. This isthe high switching level ofthe Schmitt triguer gate. Carefully turn the wiper back again until the output reverts to logic high. Tals point corre- DECEMBER 1992 {Hz MULTIFUNCTION FREQUENCY METER - 1 SYSTEM TESTS ‘Component Test Group Test n Group2 Channel B 72 Group3 SI0- Receive 13,7475 Group2 Channel B 16 Group2 Channel A Tre Group2 Gate-Out Te Group2 Channel A \C1ab Group Channel Icte Group2 Channel A icra Group2 Channel B ce test not possible icaa Group) IC3a ie3b Group 2 Channel A Ic4a Group2 Channel B Ic4b,c —Group2_pre-counter Ica Group1 cad c5 Group 2 —_pre-counter (C6,IC7 test not possibie Ice Group3 $lO-Send ice powar supply Ic10 test not possible §2-S6 Group2 keyboard B21 Group2 buzzer sponds to the low switching level of the Schmitt trigger gate. Measure the wiper voltage, add it to the high switching level and divide the result by two. Adjust the preset to give this centre voltage. Alternatively, if you do not have a high- impedance voltmeter, find the switching levels, and set the wiper as accurately as possible in between the two correspond ing wiper positions Next, the central oscillator is adjusted to 12 MHz exactly. This is best done by ap- plying a known, high-precision test fre quency to the meter, and adjusting trimmer capacitor C2 until this Frequency is displayed. In case the span of C2 is too small, fita small capacitor (20 pF) in paral lel with the quartz crystal Operation Pross the MENU key to enter the normal (measurement) mode of the instrument Menu options are selected by pressing the ‘down’ key, and the selection is confirmed by pressing the ENTER key, The next menu then appears. The EDIT key allows user defined settings to be entered. The EDIT mode can be left by pressing the ENTER key. Pressing the MENU and ENTER keys simultaneously takes you out Of the menu mode. The same key combina- tion is used to end a measurement, A more extensive description of the various menu options will be given in a future instal- mient a Part 2 of this article will give detailed descrip. tions of all ment options and instrument set- tings.DIGITAL AUDIO/VISUAL SYSTEM PART 2: CONSTRUCTION After last month's introduction and the description of the solve unit and the projector, we are now ready to tackle the construction of the various modules that make up the DiAV system. As you may recalll, there are two construction options: all modules in one enclosure, or each module in its own enclosure. In this instalment, we deal with the latter option. Design by A. Rietjens Fitting all DiAV system modules in sepa- rate enclosures has the advantage of al- lowing full use to be made of all display functions. The other option, fitting the units into a single enclosure, also has an advantage: it results in a compact piece of equipment that is easy to carry around and connect. It is also the cheaper option, since a number of displays and enclosures ‘may be omitted ‘The choice between the above two op- tions is entirely your own. Note, however, that each option is based on its own com- ponent set. This means that you first have to decide on an option, and only then start purchasing components. This will save ‘you a lot of desoldering work later, in par ticular with the connectors, This article instalment is the first of two that deals with the construction of the DiAV. As already mentioned, it will deal with the construction of the DIAV as sepa. Part 3 of the article will tackle the construction as a single unit rate modules. Part 4 will concentrate mainly on the con- trol software, Dissolve unit construction The dissolve unit described last month is built on a compact printed circuit board that allows the unit to be built into a rela- tively small case. The PCB is supplied in- clusive of the system software (sce page 110), and consists of six sub-PCBs, which have to be separated from one another be- fore they are populated. The PCB artwork is shown in Figs. 15, 16 and 17. In view of the high track density and the high num- ber of through contacts, it is not recom- mended to produce this PCB yourself Hence, the component side and solder side track layouts are shown reduced to 50% in Figs. 16 and 17. "A breaking line is fraised between the three largest sub-PCBs. This line makes the sub-PCBs easier to separate. The three other boards with Kut-Kr7, Kz0 and K21 on. them are easily separated with a jig-saw. The small piece of PCB material left over after cutting the boards should not be thrown away as it has a function later. The largest sub-PCB contains the con trol electronics, and may be assembled in the usual fashion by referring to the com ponent overlay and the parts list However, before you start soldering, you ‘may have to cut off the PCB corners near Ti and ICie. This is only necessary if you use the enclostire mentioned in the parts list. To be able to run an initial check on the board, the ICs must not be fitted as yet. ‘The same goes for connector Ki. This has to do with the Centronics output, which requires the connections of K2 to be brought out via the small board that con tains Kx» and K2i, At the side of Ka, re ‘move as much PCB material as possible to enable a flatcable connector for PCB ‘mounting to be fitted (see Fig. 18). Next, fit Kao and Kai on this board (with a 10-cm long piece of flatcable). This construction is mounted on top of Ki as illustrated in Fig. 19. To be able to do this, however, first cut off a small piece of Ki, as shown in Fig. 18. If this is not possible on your par- ticular connector, the pins of Ka must be must be cut off flush with the PCB before soldering, so that nothing protrudes from the board underside after soldering, Alternatively, if you do not intend to use the ready-made front panels supplied hrough the Readers Services, the remain- ing piece of PCB material may be inserted between the stacked connectors. Note, however, that this neither allows the eon. nector PCB to be fitted above the stacked connectors, nor in the case mentioned in the parts list. If you do not have suffi- ciently long M2.5 (2.5 mm dia.) bolts, drill out the mounting holes of the eject headers 0 that 3-mm bolts can be used instead. ‘The connector board that contains Ko Kis requires no further explanations. This board may be secured on to the main board with the aid of four 4-cm long PCB pillars (see introductory photograph and Fig. 26). To make sure that the enclosure can be closed with the top cover, the PCB pillars have to be reduced by about 2 mm. Also file little material off connectors Ki! and Ki2 to provide enough room for Ki and Kas, Mount all parts on the display board, except the LEDs. Note that the displays must be mounted on IC sockets. This is necessary to provide enough room for the LEDs behind the front panel. Before fitting the LEDs, drill and cut the front panel with the aid of the drilling template sup- ELEKTOR ELECTRONICS DECEMBER 1992122] GENERAL INTEREST COMPONENTS LIST DISSOLVE UNIT MAIN BOARD. Resistors: 8 2200, 1;R2.R7;R8; RIB:R14: R19; R20 8 47a RS.ASAS.AN; RIS;RI7R21; R23 7 1K9 R4:RIO.RI6; 22;R33:R36; a7 6 10K2. ReR12:R18; Rea:R28.R34 2 2Ka2 25:26 1 120K Rev 2 3300 29:30 2 3k03 31:32 1 47KQ. 35, 1 Bway 1k8 SIL array R38 1 100K Rao 1 10kE2 SMA (for 19; see text) 1 5OkR preset | PI Capacitors: 4 10uF 16V radial © C1-C4 1 nF os 1 1000uF 16V miniature radial C6 14 100nF c7,c12-¢24 1 150nF cs 1 4uF7 16V radial C9 2 20F 10.011 ‘Semiconductors: 9 1N4001 1 1NG148 3 BD679 75; T7TeT10;711 4 BD140 Ta:T6:T9.T12 1 BC547B 2 CNY74-4 1 0032-16 1 74HCTS73 1 2764 EPROM (886171; supplied with the PCB; see page 70) C5 4 7aHCTS74 IC6:IC8:IC9, Ic10 1 74HOTS41 \c7 1 74HOT123 lent 1 74HOT32 ler2 1 7aHCT74 le13 1 74HCTOo era 1 74HCTI39 eis 1 7805 1e16 Miscellaneous: 2 ‘4-way angled PCB header with eject latches K1:K20 3 14way boxheader —-K2iK3;K4 2 way PCB mount mini-DIN socket K5iK6 2 26-way boxheader ——K7:K13 1 2way 5-mm raster PCB terminal block KB 4 B-way 240° PCB mount DIN socket Ko-K12 1 14-way IDC header for PCB mounting kai 1 4-way angled DIP switch $1 1. chassis mount SPST switch s2 1. SPST PCB mount slide switch; angled pins $3 1 16MHz quartz crystal Xt 1. PCB mount mains adaptor socket (2.1 mm dia centre pin) 4 10cm 14-way flatcable 30cm 26-way flatcable 1 14-way IDC socket 2 26-way IDC socket 1 25-way male IDC style sub-D connector 12VAC @1A (min.) mains adaptor Elbox RE-2 enclosure (Retex)* PCB pillar; length 4em Printed circuit board and software order code 920022 (see page 110) 1. Front panel foil 920022-F1 (see page 110) 1 Rear panel foil 920022-F2 (see page 110) REMOTE CONTROL 3 push-button with change-over contact 1 G-way mini-DIN plug approx. 2m 3-wire cable DISPLAY BOARD. Resistors: 14 2200, 12 2Kn2 Capacitors 4 100nF ‘Semiconductors: 20 LED; red; rectangular; 10-014; face 5x2.5mm 4 LED; red; square; face 56mm 15,021,027; 33 4 BC5578 Tiet19 7 BC547B T4715; 720-725 1 7404543 1c18 1 T4HCT238 1e17 8 HD11070 (orange) LD1-LD8 Miscellaneous: 2 14-way PCB mount IDC header Ki8K19 2 10-0m 14-way flatoable 2 \4-way IDC socket *Retex, Jerusalén 10, 08902 Hospital, Barcelona, Spain. Tel. +34:3.335 5562 Fax 434 3.335 7468, Distributor. Boss Industrial Mouldings Litd., James Carter Road, Mildenhall Suffolk \P28 78D. Tel. (0638) 716101 plied as a 1:1 copy with the front panel foil. The drilling template is also shown (reduced) in Fig, 21. Stick the template on to the aluminium front panel, and mark all corner points with a centre punch. Next, mark out the clearances with the aid of a ruler and a scribe. The Retex case has a very ‘soft’ aluminium front panel, which is easy to cut and finish with a jig saw and a set of small files After making the holes, secure the dis play board to the front panel using PCB pillars and bolts with countersunk heads, ‘The distance between the display board and the front panel should be tuned such that the display fronts are flush with the front panel. Next, insert the LEDs into the holes, adjust their positions, and solder Fit two 10-cm long pieces of flatcable to Kis and Kis. These nected to Ks and Ks respectively via ap- bles are later con propriate connectors. Mind you: Kis and Fig. 18. llustrating how K20 and K24 and the flatcable are combined on the small conneo- tor board. Before fitting this board and K1 on to the main board, a small part has to be cut off Kt. Fig. 19. Seen here are connectors K1 and 20 fitted neatly on to the main board. LEKTOR ELECTRONICS DECEMBER 1992Ki» must be fitted at the solder side of the board, as indicated by the dashed outlines ‘on the component overlay. Before sticking the foil on to the front panel, darken the area around the LED holes (at the inside of the foil) using a wa- terproof black marker pen. This sary because the front panel foil isa little too translucent, and its design is such that only sections of the LED faces are really visible atthe outside. All clear? Then care- fully stick on the front panel foil Connection and test Before we can run our first test on the dis solve unit, it is necessary to make a small hardware modification. During the devel- opment of the system it appeared that some mains adaptors cause problems with the zero crossing detector. To be more pre cise: the mains zero crossing instants were not sufficiently accurately defined. This problem was traced down to a floating, base of T13 during the zero crossing. The remedy is simple: fit a 10-kA resistor be- tween the base and the emitter of T13 (an SMA resistor fits exactly between the base and emitter solder spots on the board). To be able to hook up the connector board, fit two IDC Ginsulation displa ment connector) sockets and one male sub-D25 connector on to a length of flatea- ble. The dimensions of this cable are shown below in Fig. 22, and the practical appearance to the right in Fig. 23. The cable at the left in Fig. 23 and the dimen- sions shown at the top in Fig, 22 are ap- plicable if the DiAV system is fitted in a single enclosure (further details in Part 3). The D25 connector may be used where the projectors are not close to the dissolve tunit In that case, one cable is sufficient to hook up all four projectors. The connec- tions made by this cable are shown in Fig. 24. One of the headers is connected to Kz, the other to Kis, Note: pin 1 goes to pin 1 (the headers and sockets are polar- ized, and havea ‘pin 1’ mark in the form of an arrow or a dot). Next, itis time to fit the first three ICs fon the ‘main board: ICi, IC2 and Cre. ‘Connect the mains adaptor to the circuit, and close switch S2. Check the presence of the supply voltage. If this is okay, switch off, open Sz, and connect projector 1 to Ke. Switch on again and check the supply volt age. If this is still okay, you are ready to test the projector connections. This is done by screwing a wire into Ks, at the ground side. The other end of the wire is used as a probe to test the carrier control input of projector 1, Touch pin 18 or pin 19 of the IC socket in position ICs with the wire end, The projector should respond by doing a slide change. Touch pin 3 of the IC socket in position IC3, whereupon the amp should light. If this works so far, the other projector connections may be tested by hooking them up to Kio, Ki and Ki, and connect- ELEKTOR E “TRONI s bl TAL AUDIO/VISUAL SYSTEM - js» 4 20. Keep to these cutting able to use the ready-made st Services (see page 110) panel, and you idhesive front panel foils supplied through the Readers 7 Zz DIAV DISSOLVE UNIT a a Fig. 21, Front panel and rear panel fol layouts (shown at about 60% of true size) MBER 1992a GENERAL INTEREST Le Fig. 22. Basic connections of the two IDC sockets and the D25 connector on the flatea~ ble. The cable lengths given at the top and the bottom of the diagram apply to ‘single enclosure’ and ‘individual enclosure’ con- struction respectively COMPONENTS LIST ‘TRIAC MODULE ‘710263 plus insulation material 12-way male RTG-22 plug ‘6-way 240" DIN socket Heat-sink SK182/37.5SA Diecast case 91%38.25mm; e.g. Hammond 15908. 6-TO-6-WAY CONNECTING CABLE 2 G-way 240° DIN plug ‘Approx. 75cm long piece of 6-way cable 6-T0-10-WAY CONNECTING CABLE 1 Bway 240° DIN plug 1 10-way DIN plug (e.g. Hirschmann ‘Type MIS100, order code 931 876-117) ‘Approx. 75cm long piece of 6-way cable ing the associated pins on IC socket ICs {pins 12-17) and IC; (pins 4, 5 and 6) to ‘ground. If these tests check out, fit the re maining components on to the main board, and comnect the display ‘After switching on, each display pair should indicate ‘01’. This indicates a cor- rect power-up, and that zero-crossings have been detected. If no alternating volt- age is applied, the decimal points on the displays will flash to indicate the error Fig. 23. Flatcables for the sing} of the DIAV system, condition, If this happens, check that you have connected an alternating voltage source, Remember, the lamp dimmer cit cuitry can not function without zero cross ings, even though the lamp intensity display gives a normal indication. This condition may be simulated on a correctly functioning board by shorting out Dy. Switch $5 allows you to choose between, two display intensity levels, The lower level ensures that the display is dimmed such that it is visible, but not obteusive, in the dark, ‘Assuming that everything checks out so far, the mininyum projector lamp inten- sity may be set by adjusting preset Pi. The lamp filaments should just glow, without producing a visible image on the projector sereen. If this can not be achieved, de- crease the value of R27 to 68 KS. Do not forget to select between ‘one-button’ (JP) closed) and “two-button’ (Pi open) projec tor control. Next, connect the remote control as shown in Fig, 25. The three switches are wired such that they keep the two micro- controller inputs ‘low’ when they are not pressed, This allows the system to detect whether the interface is controlled via the Centronics input or manually. In the latter case, the number of projectors. selected equals that set on switch block $1 (switches St and Si-1). The lamp intensity displays of the non-used projectors are au tomatically switched off, which allows the umber of used projectors to be seen at a glance. When the ‘forward’ button (Si) is pressed, the first projector will be enabled. The next button action will result in a fade from projector 1 to projector 2. Next, pro: closure’ (left) and “individual enclosures’ (right) options jector 1 will do one forward change. In this way, all connected projectors may be con- trolled one after another. On pressing the ‘reverse’ button, the previous projector will first do a reverse change (carrier one position backwards) Next, a fade is done to the previous projec tor. This simple sequential control allows the dissolve unit to be used on its own, —" g Fig. 24. The 25-way male sub-D connector a- lows all projector connections to be joined into one cable. EKTOR ELECTRONICS DECEMBER 1992SITAL AUDIO/VISUAL SYSTEM - solder sie oF sa0022 1-20 Fig. 25. Manual control: connections of the three switches to the mini-DIN plug inserted Into the I2¢ control input. with two to four projectors, where the slides are distributed over the projectors. Switch Ss serves to set the dissolve (fade ‘out/fade in) time. Pressing it once causes a LED bar to appear on all four lamp indica- tors, The LED bar indication is propor- Honal to the dissolve time. The ‘forward’ and ‘reverse’ button then serve to increase ‘and decrease the dissolve time respec tively, in steps of one second. The LED bar will change accordingly. Pressing $4 again takes you back to the projector control mode. Provision has been made for the projec tor illumination areas (on the screen) to be matched. All four projectors light when the Torward’ of ‘reverse’ button is pressed when the unit is switched on. This allows you to position the projectors such that their light beams overlap exactly on the screen, This mode is left by pressing the Fig. 26. The side panel of the Retex enclosure has holes for the mains adaptor plug, the switches and the potentiometer. The case If everything works to your satisfaction, the dissolve unit may be fitted into its en- closure. SelFadhesive foils are available for the front as well as the rear panel of the Retex enclosure. The rear panel layout is shown reduced in Fig. 21. These foils give the unit a professional and attractive fin ish, There is one point to note about the rear panel and the rear panel foil: in some cases, the hole for the mini-DIN plug may have to be made larger than indicated by The righthand side panel of the Retex cease is drilled and filed to allow Si, $2, $5 and P1 to be operated. Also note the hote required to insert the mains adaptor plug (see Fig. 26). The voltage regulator is bolted on to the rear panel. If you use an- other enclosure than the one we recom: mend, remember that the metal part of ICisand the sides of the IFC plugs are con- nected to ground, To prevent a short cit- cuit with the alternating voltage, the ‘mains adaptor input socket must be an in- sulated type ifitis fitted on the same panel ‘forward’ or ‘reverse’ button again, where- the drilling template. This is necessary be- as the regulator. a upon the system can be used as described cause the plug body has to touch the above socket to make proper contact. 1 a 1 8751 Emulator March 1992, p. 53. (Corrections; component information) While in emulation mode, the register con- tents are displayed with an offset of one vertical line from the associated register dlesignations. This error occurs on early re- leases of the system software, item ESS 17H1, and is caused by one superfluous ‘space’ character in the DEV.EXE program. This ‘space’ (20H) should be changed into a ‘line feed’ (OAH), First, make a backup copy of your original diskette. Next, use a hex editor to change the byte at address offset DEODH from 20H into OAH. Using the hex editor of PCTools V6, for instance, this byte is found in relative sector 111 (decimal!) at offset ODE. 53 and 54 of the SC80C451 must be connected to ground to give proper access to (simulated) Port 0, For no apparent reason, this is not indicated in the Signetics datasheets. Port 0 is actually CORRECTIONS AND UPDATES simulated by Port 6 of the SCS0C4S1. For further information an this compatibility problem with generic 8051 assembler files, consult the SCSOCIS1 (Signetics) or 8xC451 (Intel) datasheets, In addition to your local Signetics (Philips Semiconductors) distributors, two suggested suppliers of the controller Type SCSOCASICCN64 are (2) Macro Marketing Ltd., Burnham Lane, Slough SLI 6LN. Telephone (0628) 604388, (@) C-1 Electronics, P.O. Box 22089, 6360 AB Nath, Holland. Fax: +31 45 241877. ELEKTOR ELECTRONICS DECEMBER 1992 GAL programmer May 1992, p. 35 (Update) The transistors Type BC369 in positions TS and 77 are apparently difficult to obtain and may be replaced by BCOd0s “The most racent version ofthe software is V. 6.5341, June 1992. The README file contains an update note on problems with the programming of certain GAL makes, as well as a suggestion to make GALs with 4 damaged electronic signature type den- tier) useab 8051 Single board computer ‘October 1992, p. 40 (Update) Since the publication of this article, we have been advised that the telephone number of Suncoast Technologies is +1 (904) 596-7599, againUNBLOCKING THE PUMP By Bryan Hart, BSc, C.Eng., MIEE Although the pump circuit, in one or other of its many guises, has been used in electronics for some fifty years, textbook explanations of its operation tend to be either sketchy and qualitative or overly mathematical. This article aims to remove the confusion surrounding the circuit by concentrating on its physical operating mechanisms and to show that algebraic analysis can be replaced by a simple, but novel, graphical procedure. 'n use in electronics for over fiffty years, the pump circuits has appeared in a wide variety of applications that include count- ing/frequency division; frequency-voltage conversion: frequency-sensitive switch de- sign demodulation; and stairease voltage gen- eration for display systems. Unfortunately, textbook treatment of its operation tend (0 be either sketchy and qualitative or overly ‘mathematical. Budding engineers studying the circuit for the first time might be for- aiven the resulting mental block The principal aims of this article are to remove that confusion by concentrating on physical operating mechanisms and to show that algebraic analysis can be replaced by @ simple, but novel, graphical procedure, This in showing the output voltage as function of acharge increment transterred daring an input pulse. The result is an eas: ily constructed "U,/AQ lattice plot” from ‘which the output waveform can be sketched by inspection, Pump circuit modelling A basic form of the pump circuit is shown in Fig. 1. For reasons that will become ap- parent Later, itis sometimes more colour: Tully known as a “cup-and-bueker' cirewit Fig. 1. Basic diode pump circuit In modelling the circuit, itis desirable first to idealize the properties of the compo- nents used. Then, when the general operat ing principles are understood, the conse- quences of using non-ideal components are ‘more easily appreciated. Thus, Cand C, have no defects such as leakage resistance that require parasiticelements to model them. The pulse generator, PG., has a constant output resistance, R; and produces an out: put waveform as shown in Fig, 2a that is a train of rectangular pulses, P,P ..Py Py having zero transition times. pulse duration iy, pulse recurrence time 7. and open-circuit pulse amplitude & Fig. 2a. Assumed input waveform. The P.G. is modelled in Fig. 2b by an ideal switch Swe and battery E. Initially Sivg is at position *a’ and between pulses, and at °b’ when the pulses are present, The time taken to switch from "a" to"b and vice versa is taken as zero, Fig. 2b. Equivalent representation by a ‘mechanical switch. Diodes D, and D, ure also modelled by ideal switehes as shown in Fig. 3. In the ‘composite circuit model in Fig. 4, the switches Sw.S5w/and Swoareall ganged together. The overall effet produced in the circuit by the repeated movement hackward and forwardof the moving parts of the switches is analo- {gous fo that produced in a hydraulic system byamechanically operated piston pump dis- Fig. 3. dealized electrical description of D, and D,. placing a quantity of Muid, so the adjective “pump” used to describe the circuit is quite appropriate, In our case, of course, the “fluid pumped’ is electric charge. Circuit operation and graphical development Prior to the arsival of the first pulse, P,, pacitors C; and C; in Fig. 4 are both un- charged. This condition, while not obvious for C,, is guaranteed by a voltage level sens- ‘ngcircuit connectedacross it, Since that piece of circuitry is not relevant at present, a brief discussion ofits left tothe end of the article. Afier the arrival of P, (‘the first stroke of the pump’), D, cuts off, D, switches on and the reduced equivalent circuit is shown in Fig. Sa. This time constant switeh- tL 4 Fig. 4, Modelling the pump with switches. ELEKTOR ELECTRONICS DECEMBER 1992Fig, 59. Equivalent circuit during P, Fig. 5b. Circuit response to P, for 1P5OR, (C=E.CAC+C). F Fig, Se. Theoretical response for Ry=0. ing eireuit, the time constant being the prod- uct CR. In this. CaC\CM(C;+C,) and is the clfective capacitance of C, and C; in series, (On the leading edge of P,, the whole of the input voltage, E, appears acorss Ry, be- ceause C, and Care initially uncharged and the potential differences across their plates ‘cannot change instantaneously. The initial charging current is therefore £/R, and it de cays exponentially to zero as C, and C; are charged. The charging current causes an equal increment of charge to be deposited (on the plates of both capacitors. The time elapsing between the 10% and 90% output voltage levels is 2.2CR, and, forall practical purposes, the charging pro- cess is taken as complete in a time interval SCR;, Let AQ, be the charge increment rans- ferred from E to Cand C; during pulse P) Provided 1>5CR;, the ease shown in Fig. Sb, wwe can write AQ\=CE, nH ELEKTOR ELECTRONICS DECEMBER 1992 [Note that the symbol A. rather than 8, is em- ‘change is not necessar- [1] deserves a further brief dis- cussion. Subject to the condition placed on Rg, itmeans that AQ, is independent of Ry ‘Thereasonisthat by thetime the railingedge of P, appears the potential difference across R, is zero. Hence, the full applied exm.l. E appears across the equivalent capacitor C. ‘The output voltage, U when Phas passed, tnd the first output “step” AU, are given by Uj AU =AQIC EC ENC HC). Ry The condition 1>5CRg, which we will assume 10 be valid from now on, is easily met in practice. Thus, if Rj=5002 (a typical vvalue)and C=).1 uF the conditionis 4225 ps. Figure Se corresponds to the case Ry=0, Which is implicit in some textbook discus- sions. However, it is purely an abstraction. Itsequires an understanding of the more dif- ficult mathematical concept of current im= pulse of infinite amplitude and zero dura~ tion, but nevertheless finite area, that dumps a charge AQ, on Cand Cin zero time. [Lis best to regard the situation shown in Fig. Se as the theoretical limit ease of what would. happen in Fig. Sb if Re were made progres- sively smaller ‘A related method of finding step ampli- tude leads on 10 a graphical procedure that is best described below, The charge traps- ferred from Eto, during P\is¢ that transferred to C; is, of course, the same and is given by AQ=C.AU Equating these two expressions for AQ, again produces eq. [2]-The same result is also obtained trom a *U,/A@ plot’. On this. lines are drawn as if C) were part of a volt ‘age source withe.m.f.£,and C, were the load To see how the plot arises, refer back 10 Fig. Sa "Taking into account voltage drops a the tend of P,, the "source characteristic for P, sccel» for short, of E and Cis Fig. 6. A“UJ40 plot for finding 40, and AU at the end of P, Bold lines show souree ‘and load characteristics. UNBLOCKING THE PUMP U=E-AOIC). BI Similarly, the “load charaeteristic’ for Py. hesb.of G U=AQIC), 141 Inthese two expressions, the number sub- scripts for U,and AQ have been omitted be~ cause these quantities are now regarded as ‘variables whose values are to be found. Consider, now, the L/AQ plot of Fis. 6, where s.c.cl» is @ straight Tine with slope =(1/C)) that passes through the axes points D(UjsE)andA(AQ=CE); Le isastraight Tine with slope-+(1/Cy)that passes through the origin and which is most easily plotted by Tocating convenient point on it other than the origin. For C, expressed in gl. a suitable point is U=1 V, AQ=Cyx1) uC ‘The intersection point, A,.ofs.c.<1> and Le. doccurs where AQ(=AQ,.) is the same charge inerement transferred to both ca pacitors. U,(=AU)) can be oblained by inspection. Fig. 7, Circuit conditions after P, and before P,: (a) C, discharging - /.=dis- charge current; (b) C discharged ~ i.=0. At the end of pulse P), D: cuts off and D, switcheson. The chargeon C,remainsal AQ. but that on C; decays to zero with time con stant C Rc—see Fig. 7a, Since this decay re quires a time interval SC,R% to complete, a second condition placed on (is((45C\Ro), but has a U, axis intercept AU, ‘On the U/AQ plot of Fig. 9, intersec- tion point A: gives AQ,, Us. and hence AU, ‘The plotting procedure for further pulses. P, ... and so on, now becomes clear. The source characteristcremainsfixed:itiss.c.<} However, there is a new load characteristic foreach successive pulse. Ths is parallel to that for the preceding pulse, butisshifted ver- tically up the U, axis by an amount equal to the previous step. The overall result is an easily constructed “lattice diagram”. From it, the output waveform can be readily ob- twined by cross-ploting asin Fig. 9b, In Fig. 10, 0. and OA, are the load characteristics respectively for P, and P,.. From the geometry of the figure m7 wen Fig. 9b. Cross-plot to show output waveform (bold line) and input waveform (faint line). Fig. 11. Modification of the source characteristic for equal steps. ‘Thus, the amplitudes of the output steps form a geometrical progression. By inspec: tion of the plot, the maximum value of U, is E, which is reached after an infinite number ‘of input pulses. We can look on C) asacup, Casa eylin- rical bucket and Esa tank of water. Then, as far as the output is concemed, the opera tion of the diode pump is analogous to that ‘of using the cup 10 scoop up water trom the tank and dump it into the bucket. In the first scoop, the cup is full, ut the next time round itis only a fraction K full; the third time. it is a fraction K" full, and so on. As a result the water level rises up the inside of the bucket by decreasing amounts after each dumping, Step equalization In an instrumentation application, such as the display of the terminal characteristics of a semiconductor device on a ‘Curve tracer’, we normally require the output voltage steps in Fig. 9b to be equal. Using ‘our water analogy. the ‘cup’ must be full for each dumping. ‘The graphical method described provides, Fig. 10. Construction for finding the ratio of magnitudes of ‘successive steps. Fig. 12. Equivalent circuit interpreation of Fig. 11.Fig. 13. ‘Bootstrap’ step equalization circuit. UNBLOCKING THE Fel Fig. 14. Charge-transfer circuit requirement for fixed ‘source characteristic and 1@,= : Bs Fig. 15. U/.1@ plot for Fig. 14. 4 logical approach to the solution of the problem of how to modify the basic circuit to obtain the required result. Looking at Fig, 9a, it follows that for ‘equal steps the circuit must be changed sothat the intersection points A\...A, are equidis- tant from the U, axis. This is possible if the Vertical shift in the load characteristic asso. ciated with a given pulse is matched by an «equal vertical shiltinthe corresponding source characteristic. As indicated in Fig. 11. this means having s.c.:2s located a distance AU, above s.c.«1>, and So on. To produce this ef fect, C, must be charged up to the output voltage on the trailing edge of each pulse. Figure 12 shows the required equivalent cir- cuit and Fig, 13 gives a hardware imple- mentation, the ‘bootstrap’ scheme. In this, a high inputimpedance opamp. Ais strapped asa voltage-fllower,thelow output impedance ‘of which is suitable not only for driving an external load, but also for supplying the ‘pre- charge” current, iy of C, {An alternative solution is to provide cir- couitry that keeps the source characteristic fixed. but, at the same time, facilitates the extrac: ionof the charge increments passing through, Ds. This can be achieved if the cathode of D; is held at a constant potential, a sensible choice for which is “circuit common *. The required circuit must, therefore, be able 10 performthe function indicatedin Fig. 14, The related U/AQ plotisin Fig, 15. Fromthe view- point of the load, the source characteristic e TOK ELECTRONICS DE BER 1992 Fig. 16. ‘Miller’ step equalization scheme. D— Fig. 17. Block schematic of reset circuit suitable tor use with circuits of Fig. 1 and Fig. 13. Reverse the polarity of the transistor for use with Fig. 16. D/A, appears to be a vertical line through A corresponding to. AQ=CE. The load charac: teristics are now required to have a slope AUC) ‘An elegant hardware implementation of this is the “Miller” scheme of Fig. 16, The cathode of D. is held at eircuil-common po- tential by the feedback action of the invert- ing opamp configuration. Hence, Reset circuit ‘As mentioned in the beginning, ancillary circuitry is required to set the initial cond tions, In practice, this means the use of a reset arrangement such as that shown in block-schematic form in Fig. 17. A Schmitt trigger senses the output voltage and trig ‘gersa monostable, M.S., when a preset level, determined by U, is reached. The output pulse of the monostable drives a discharge transistor connected across Cy Concluding comments ‘This article has described the development ‘of agraphical procedure for sketching, rapidly the output voltage waveformofadiode pump citeuit driven by a train of rectangular input pulses. It may at first seem odd 10 be plot ting voltage versus charge because it is so rarely done other than in, perhaps, early physics laboratory work with capacitors: nevertheless, nothing could more nearly de- scribe the essential behaviour of a capacitor The plotting procedure itself does not necessatily require a knowledge of the sm amountofalgebrathat hasbeen includedhere to justify the method. Acknowledegments ‘Thanks are due to Mr RH Pearson for inc0 COMPRESSOR/LIMITER she compressor is based on two series-connected attenuator networks, whose attenuation is controlled by light-dependent re sistors(LDR9) that are illuminated by light-emitting diodes (LEDs) ‘The input signal is applied to the non-inverting input of opamp IC), via Rj. Chroust IC, sn con: junction with D3, Dy and ICyy, provides full-wave rectification of the signal. The resulting direct voltage is used to charge C5 via Ds. The diode allows fast charg. ing of the capacitor, which can discharge only via R; ‘Compression proper is provided by IC\¢. Depending on the setting, of P; and P2, the output voltage of ICje. drops when the potential across C5 reaches a certain value. This causes Tz and, via ICjq. T1 light and the input signal is at zy T” tenuated. Lil The attack time of the circuit r is determined by the speed of the - i mer! We basoet = The amplification of IC. and > =. eat The cireuit as shown acts as a lim- Thisrangecanbeextended byaddingone determined largely by the LEDs and is ter: ifthe #ve Input of IC), 18 connected or more attenuator sections or by in- 50 mA maximum, (o the output of the circuit, a standard creasing the value of Ry and Ro. compressor is obtained. ‘The LDR/LED combinations must be (J. Barendrecht - 924096) MINIATURE CRYSTAL OSCILLATOR ‘owadays, with the advent of SMD (surface mount design}, it is possible to construct tiny circuits. In the case of crystal oscillators that may, however, not IC, the maximum frequency was found to be about 8 MHz. Finally, the eircult works per- fectly all ight with standard com: ponents as well always be feasible because of the size of the crystal. Fortunately Statek, a specialist crystal man. uufacturer, produces crystals mea suring only 8\4s<1 mm for SMD. ‘Together with a single inverter and four passive components, such a crystal makes it possible to make a truly m ble oscillator as show The cireuit works very wellup. to frequencies of 16 MHz if an HC.IC ts used. With an HCT (Statek Application - 924113) let = 7aHcos ena ELEKTOR ELECTRONICS DECEMBER 1992VIDEO DEMODULATOR 1e demodulator is based on a Type TDA8S41 chip, which is the suc- cessor of the well-known Types TDA2541 and TDAS541. Apart from a demodu lator. the chip contains an AGC fauto- matic gain control) section for tuners whose AGC voltage is directly propor- Uonal (o the gain, and an AFC (auto- ‘matte frequency control) facility. In the present design, the AFC is used merely as a tuning indicator. ‘The control range ofthe AGC is 67 €B. Because of internal supply regulation, the input sensitivity of the IC is 40 pV. ‘The IC provides a video signal at a level of 2.7 V at pin 12, from where it is passed to a Toko low-pass filter, Fly which removes any residual carrier fre~ quencies from the video signal. The fil: ler has an attenuation of 6 dB over its pass-band. To obtain a standard video signal of 1 Vpp into 75 0, the filter may w av roweLmu-rsisaco ELERTOR ELECTRONICS CEMBER 1992 be followed by a video amplifier/butfer, for instance, a Type NE592. Note that the output impedance of the filteris 1k. ‘The AGC operating point is set with P). The AGC control voltage is taken from pin 4 via Rp. The level of the cur- rent through the AGC output is limited to about 10 mA. Network R7-Ci1-Ci2 forms an AGC detector, which also pro- vides pulses fora sample & hold eireuit ‘That circuit ensures that no video infor mation is present in the AFC output, ‘The reservoir capacitor for the S&H ctr cuit is Co. If pin 6 is connected directly to earth, the AFC is disabled and the po- tential at pin 5 is then roughly half the supply voltage. The AFC requires a synchronous de- modulator with its own tuned circuit LCs, Because of parasitic) capacitive cou pling (Co-C 9) with the tuned etreust for the reference amplifier, Ly-C7. the skirt of the AFC characteristic becomes steeper. Since the AFC voltage here provides a tuning indication. a steep characteristic is not really desirable, so that Cy and Cp must be kept as small as possible. The tuning indicator is formed by centre-zero meter My. Tuning the reference circuit (to re. move any residue of the carrier} is tricky and can really be done properly only if an rf. analyser or modulated rf. gen erator and frequency meter are avail able. If crystal-controlled PLL (phase locked loop) tuningis used, however. the adjustment can be carried out without this test equipment. Once the channel tuning and the ad- justment of Ly-Czarecorreet, the ARC cir- cuit, Lo-Cs, may be tuned for centre- zeroreadingof the meter (half supply volt age at the AFC output). Resistors Rs and Rg hold the other terminal of M, at half the supply voltage and at the same time limit the current through the meter to about 100 yA. Because of the circuit layout, there may be cross-talk between the video output and the i. input. This may, however, be cured by connecting a 6.8 iH choke in series with pin 12 (as close to the IC as Feasible). ‘The demodulator draws a current of about 45 mA. ‘The input signal may be taken from a SAW (surface acoustic wave) filter which is readily available nowadays Most tuners—although this must be checked—are capable of driving a vari- ely of SAW filters. (7. Giesberts - 924078)SOLID STATE T/R TRANSVERTER INTERFACE This little circuit is intended for UHF transverters in com ation with a 2-m band transceiver. It is simple to build and much cheaper than a coax relay. Design by Pedro Wyns, ON4AWQ, Athen with the use of a 70m or Sem transverter in combination with a 2. n transceiver is that the former nearly always has two IF (intermediate frequency) connections: the transmitter input and the receiver output. By contrast, the 2-m rig has only one RF connection. But there are more pitfalls: usually, the transverter can not cope with the normal ‘output power of the transceiver, so that some attenuation isin order. Second, some means has to be devised to enable the transverter to switch between receiving and transmitting under the control of the 2-m transceiver. In most cases, this means that the transceiver has to be opened to bring out a transmit/receive switching voltage that can be used to energize a coax relay. All of these problems may be over: come one way or another, and radio ama- teurs are not the most faint hearted of electronics enthusiasts. However, one of the most awkward problems tied up with, ‘getting on the air on 70 or 23 is not strictly speaking, a technical one: itis the cost ofa suitable coax relay at the input of the transverter (well, yes, there may be the ‘odd technicality to sort out with the YL or XYL regarding finance matters, but these will not be gone into here), The circuit shown in Fig. 1 is an all- solid-state equivalent of an expensive coax relay at the transceiver side of a 70-cm or 23-cm transverter. Evidently, a coax relay is still required at the output of the trans- verter, nothing we can do about that! The operation of the circuit is fairly simple The RX (receive) and TX (transmit) supply voltages of the transverter are brought under the control of the PTT (push-to-talk) switch of the 2m transceiver. This is ‘hieved with a pair of complementary (npn/pnp) medium-power transistors Type BD139/BD140. In receive mode, the tansverter output snal arrives at the transceiver input via two pieces of coax, Zz and Zi, When the 2-m transceiver starts to transmit, the +TX transverter supply line switches to +12 V. The resulting direct, voltage that arrives on the PIN diodes causes the transceiver’s RF output energy to be fed to a pi-type attenuator consisting cf three resistors. Also, the two quart wave pieces of coax, Zi and Z2, then form a notch (band-stop filter) for the 2-m sig. nal, so that virtually no RF signal arrives at the transverter’s receiver output. The two pairs of anti-parallel PIN diodes also protect the transverter’s receiver output against RF power when the supply voltage happens to be off The value and the power rating of the resistors in the pi-attenuator, R7-R&-Re, de pend on the degree of attenuation re- quired, and must, therefore, be calculated fon the basis of the transceiver output power and the transverter input power. The RF losses introduced by the PIN diodes are negligible since the switch isin serted at the IF (intermediate frequency) output of the transverter, As is well known from UHF and SHE receiver engi neering, it is the input stage, not the IF stage, of a receiver that determines the overall noise figure (provided there is suf- fic conversion gain), ‘The transverter interface should not be too difficult to build using ‘dead bug’ tech niques familiar from experimental RF con structing. The transistor pait is conveniently accommodated on a piece of stripboard ‘The length of the coax pieces is about 0,66 times the quarter-wave length of the IF signal, Ifa 2-m rig is used, the length is = 300/145 MHz (A= 50cm length = 0.66 » 50 cm = 33 cm The factor 0.66 is the so-called velocity fac- tor of the coax cable, and applies to most types of 50-2 cable. The four cable ends must be grounded via the shielding braid The author used RGSS coax for the proto- type, although other cables may be used as well, for instance, RGI74U, which is much thinner, Teflon (PTEE) coax is even better, but check the velocity factor to make sure that the physical lengths of Zi and Z2 are correct. Finally, the +12V Rx and +12V Tx outputs of the circuit are capable of sup- plying up to about 300 mA. Where higher currents are required, Ti and T2 must be replaced with power darlingtons. . iv GND to PTT 2m Taev “. set the wanted peak value of the supply to the sensor. In many cases, a peak value of 1 V will be found suitable: the output voltage then varies between 0 Vand 5 V. (0. Ruiters ~ 924086) Figaro produets are av lable from USA (Head Office): Figaro USA Ine PO. Box 357 1000 Skokie Boulevard Room 575 Wilmette HL 60091 USA Telephone 0708 256 3546 Fax 0708 256 3884 United Kingdom: Envin Sciemific Products Lid 28 High Street Aylburton nr Lydney Gloucester GLIS SDA. ‘Telephone (0594) 844 707 Fax (0594) 844 722 Europe. Figaro Inc. Oststrasse 10 1D-4000 Dusseldorf Germany Telephone +49 211 358 128 Fax +49 211 359 538 ELEKTOR ELECTRONICS DECEMBER 199240 W OUTPUT AMPLIFIER Aittioush there are a number of hy prid output modules on the market. very few of them combine compactness with reasonable price and good perfor mance. One of these few is SGS's tsed in the present amplifier. ‘The design of the amplifier is straight forward: a power opamp followed by two output transistors. The audio signal is applied to the non inverting input of power opamp IC, via socket K; and capacitor Cj. The supply current to the IC varies in accordance with the input signal. Consequently, there will bea similarly varying voltage drop across resistors Rg, Ry, Rg, and Ra since these are in the supply lines to the opamp. As long as the current is lower than about 1 A, the voltage drop across the resistors will be insufficient to switch on transistors T; and Tz. This means that outputs up to 2 W into 4 Q are pro: vided entirely by the opamp. Once the out. putcurrentexceedsa levelof lA. sistorsare onand contribute tothe power output ge5e Byer, B was ojffo Sg care ont ah Perce ol oe f ° 83, When the input signal is small, there fsno quiescent current through the tran. sistor, but there is through the opamp. Crossover problems are thus obviated Since the IC also provides thermal com. pensation, stability of the operating point is ensured. The supply voltage may lie between 12V and an absolute maximum of 44 V. Construction of the amplifier on the printed-circutt board should be straight. forward. The transistors as well as the IC must be fitted insulated on to a heat sink of about 2 k W-1. Use plenty of heat conducting paste. ‘The supply line should be protected by a3.15, fuse. (SGS application 924054) ‘Technical Data aay. 22 Wino 82 40 W into 40 Supply voltage Maximum output (for THD=0.1%) Hamonie distortion 1 KHZ/8 QL W 0.012% 1 kHz/4 2720 W 0.032% 20KHBONIW 0.0789 2kHZ4OR0W 0.2% UKHZ/8 On W. 0.038% 1 KH2/4 ON W 0.044% Quiescent current about 38mA. | Efficiency 8a (maximum load) 4.0 Parts List Resistors: RI-R4 = 100k R5=8.2kQ RO-R9= 1.4.0. 1% RI0=19 Capacitors: 170 nF 10 uF, 63 V, radial 7 JF, 63 V, radial C4,C5,C7 Miscellaneous: K1 = audio socket Heat sink 2 K W-! Insulating washers, ete, for ICI, TH, T2 ELEKTOR ELECTRONICS DECEMBER 1992PULSE GENERATOR FOR AV RECORDERS V recorders used in audio-video pr -ntationscontain an additional head to read and write the control pulses for the slide projector. These pulses are nor: mally used io actuate a relay that oper- ates the projector. The pulse dataare usu. ally written on ane of the audio tracks at the non-used side of the tape. Since the extra head in modern recorders does not make use of the two standard audio tracks, itecomes possible tohave stereo sound for the presentation. This does mean, however, that only one side of the cassette can be used The pulse generator is a complete cir- cuit—see diagram—for utilizingthe extra head (L,) of an AV recorder. The pulses are written on the tape with the aid of S, while S; is used to select on/off record. and play. When S) is set to its centre position (play), the pulse signal is applied to amplifier T;-Ty via Cs. Ata certain signal strength the level at the collector of T; goes high, whereupon T, is switched on via Cy and Sy. The relay is thenactuated; C, ensures that it does not clatter. With S, in position 3. 6 (record), Ty is, connected to amplifier T;-T,, As long as S; is not operated. T, is on and T; is off. However. the stage based on T, causes Ty to switch off for a short time. During that time, T, is on and amplifier oscillates owing to the feedback via R; Since the emitter of Ty is linked to Lj, the oscillatory signal is written on the tape. During the remainder of the time, T and thus T,, is on and the tape ts erased ‘When §, is pressed, T, conducts for a period determined by the time-constant R,-C, (here 100 ms}. Because of this, C) is discharged rapidly. and T, is switched off, Assoonas'T} is offagain, C,is charged via R;. After about one second, the po lential across C; has risen to a level at which T, begins to conduct again. Thus, pressing S; causes a one-second pulse to be written on the tape. Because the collector of nected to T; via Ry, and Dy, th jated when S, Is pressed, irrespec lave of whether S, is in position play or record. If for a given projector the pulse du ration is too long. it may be shortened by reducing the value of Cs The ITTAV recorder shown in the pho tograph contains a ci one described here. (A. Rietjens ~ 924057) reuit similar to the ecsi7e | ecssre fecs78 Tare ELEKTOR ELECTRONICS DECEMBER 1992+6] IDC TO BOX HEADER ADAPTOR Tiescotyouwhohave ever worked with flateables will know that IDCs (insu- lation displacement connectors) are sim- ple to use, and give reliable connections. IDCs are available as sockets and plugs. and used extensively to connect flatea- bles to double-raw box headers or pin. headers on computer cards that offer just about any type of interface to the out. side world (a good example is the multi purpose Z80 card described in Ref. 1) The present adaptor boards (there are six of them contained on the PCB shown here)are, for instance. perfect for‘chang- ing’ from IDC to eject-header style con nectors fasillustrated by the photographs) when a flateable runs from a board to a connector on the rear panel of the en- closure. Also, in many cases, an adap. torboard fitted with aneject-style header will be cheaper and more flexible (when it comes to connecting and disconnect ingilateables) than a press-on (IDC style) sub-D socket or plug. ‘The spots on the component overlay indicate the position of the holes that have to be drilled when smaller types of eject header are used. Straight headers with 10, 14, 16, 18 or 20 pins may be fitted. Iyou fit two box headers on an adaptor board, you can use it to couple flatea. bles terminated into IDC sockets. In that way. you can make IDC extension ca bles, which are particularly useful when PCB with lots of flateable connections is removed from an enclosure for repatr or inspection. (A. Rietjens - 924039) Reference: 1. "Multi-purpose 280 card", Blektor Electronics, May and June 1992LOW NOISE AMPLIFIER | Oreraystdesenings iow noise am plifieris theshuntingofseveralinput stages. This reduces the overall noise. ugetrt-l'2, where Un 38 the total noise potential of n stages; uj Is the noise volt aye of one stage: and nis the number of stages, This design Is entirely feasible because noise Is a randomly composed signal. Therefore, the noise signals of a number of stages at any one moment are highly Ukely to have a diferent tre quency and phase, so that they partly neutralize one another. In the presentamplifier, three low-noise ppamps. IC,-IC3, are connected in par allel. Accordingta the manufacturers’ data sheet. thenolseofa single LT1028 amounts (0 0.9 nV Hz”, To this must be added the thermal noise generated by resistors Ro-Rj1. Cireuit IC, sums and amplifies, the output signals of IC |-ICs. Measurements on the prototype show a total noise of 0.67 nV Hz, According (o the earlier formulla, the three opamps have an overall noise of 0.52 nV Hz-!/ ‘The diflerence of 0.15 nV between this and the measured figure is caused by the resistors. This is a low figure bear- ing in mind that a 1 @ resistor at room temperature generates a thermal noise oF 0.13 nV Hei, The amplification, a, of the eireuit i computed from =n 4Ry/ RVR / Re Its necessary that the three resistors In groups Ry, Rs, and Ry: Ro, Ry, and Rg; and Rg, Ra. and Ry, have identical values. With values as Shown, the cir ‘cuit has an amplification of »600. Apart from having a low noise out put, the Type LT1028 opamp is also last: it has a slew rate of 15 Vist and a bandwidth of 75 Mitz for o=-1. Even when the amplification is »63, the band- ‘width of the elreuit as a whole, but with: out Ryg and C)7, 1s 2 MHz. However, toavoid signal overshoot, the bandwidth Js limited by Ry2-C)7 to 500 kHz, which is more than adequate for even the most demanding audio application. ‘The THD+noise ratio at a 1 kHz ant pul al a Tevel of 1 V is only 0.008%. Ifyou wish (a experiment with the T1028, bear in mind that its stability Js internally compensated for amplifica ions of >»2. Since resistors Ry, Ry, and Re contribute most lo the overall noise. iis necessary that their value is kept as ow as possible. Naturally. all resistors, used should be metal fllm types. With a supply voltage of #15 V, each IC draws a current of about 7.5 mA. (T. Giesberts - 924087) av ELERTOR ELECTRONICS DECEMBER 1992MINI KEYBOARD FOR Z-80 CARD pending on the application of the ‘Multifunction 280 card’ (Ref. 1), a basic set of switches to control the pro- gram flow may be required. Also, in some cases, the PC/XT keyboard that can be connected to the card may prove a little too bulky to be carried around. ‘The keyboard plus LED indication de- scribed here is connected to PIOAon the multi-function 280 card. The 10 keys are arranged horizontally (see PCB lay. out) on a board fitted behind the en. closure front panel, and serve to con- trol functions determined by the user software. With the exception of S1 and Sip. each key has an associated LED to indicate the key status. The number of the key pressed is con- veyed to the PIO in the form of an in- verted 4-bit binary number supplied by IC, a 10-to-4 line priority encoder Type ‘7AHCTL47. No key pressed gives output value 15 (binary), while, for instance, key 2 gives 15-213, Bits 6 and 7 are normally logic high, but can be strapped low if desired by fit- ting jumpers. They may be used as re ‘quired by the application. A suggestion: the 280 software, on reading the key number, causes the PIO on the Z80 card to pull line PAd high, which enables shift register IC). Next, the PIO causes one of the LEDs to light by outputting the ap- propriate number of elock pulses via line PAS. Since the LED activity is controlled by the keyboard decoder software, rather than directly by the keys, the user has a ‘good indication that the selected function has been accepted by the system. In principle, K; on the keyboard unit can be connected to any of the three P1Os on the Z80 card: Ky, Kg or Ks. The keyboard /LED routines in the BIOS EPROM (€S86121—sce p. 110)). however, are based on connection to Ky (PIOA). The routines ‘"READEXTRAKEY’ and ‘LED: OUTPUT provided in the BIOS, and a demonstration program on the project diskette (ESS1711—see p. 110), make life ‘easy for the 280 programmer by offering simple ways of reading the keys. and controlling the LEDs. respectively. ‘The user is, of course. free to deter- mine the functions of the keys and LEDs, Qn 0100 lL. ® @ Fe is Pada ee ELEKTOR ELECTRONICS DECEMBER 1992or, indeed, to omit one or more keys or LEDs not required for his application. Switch Sjo has a special function, land is not read by the 280: it is the LCD. back light contral switeh that uses two bistables (IC3, and ICyy) to provide a toggle function Transistor T) is effectively connected in series with the back light supply and the back light input of the LCD (more details on this may be found in Ref. 2) The outputs marked "BL' are connected to the jumper marked 'LCD’on the 280 card. If the back light is very weak, the SL connections should be inierehanged. In this application, the BD 140 will hap- pily work in both direetions, but its eur rent gain will be much lower when the cal: lector functions as the emitter. Construction of the keyboard unit is simple. As indicated by the dashed out- lineson the component overlay. the keys and LEDs are fitted at the solder side of the board. leo00e 88000 Se000 sec00 Seoco eeoce S8e0o a ee © $6600\0% 6 Ooo 0 e MINI KEYBOARD POR 7-80 CARD Finally, the unit is connected to the 280 card viaa length of 14-way flat-cable Atted with IDC sackets. Connector Ky may be either a 14-way box header or a L-way male IDC connector. The lat- ter may be soldered permanentiy to the PCB. if the other end of the cable is fit ted with an IDC socket. Alternatively ia box header {s used in position Ky, the flateable will have IDC sockets at both ends. Power is supplied via this cable by the Z80 card. (A, Rieyens - 924047) Reference: 1. “Multipurpose 780 card” Electronics May and June 1992. Blektor IOAN R12 RI4IR16-R23 = 10K02 Ris = 3900 Capacit C1:C2;03 = 100nF ‘Semiconductors: D1-D8 = LED, 3mm, red Ic2=74HOT147 |IC3=74HCT74 Miscellaneous: | Ki = 14-way box header or male !DC connector (see tex!) $1-S10 = PCB-mount push-to-make but ton, Type 3CTLS (Amrah), PCB Type 924047. ATOR ELECTRONICS DECEMBER 1992LOW-DROP DIODE Aiiigtcarrents, many stcon diodes have a forward voltage of 1 V or more. ‘There are types whose drop at currents of up to 2-3 A is limited (0 0.5-0.6 V. but even that may cause unacceptably high losses. The circuit described here of- fers a possible remedy. ‘The cathode. C, of Ty. a SinMos Fer, 18 connected toa sinusoidal-voltage source. ‘The anode, A. thus functions as a refer ence point. Capacitor C, is charged to the peak value of the sinusoidal voltage, U,, via Dy. This ensures that the opamp is provided with power even during the negative half-periods of U,. ‘The non-inverting input of IC, is set to half the peak value of Ui, via potential divider R,-P)-Ry. Because of voltage di- vider Rp-Ry, the potential at the invert inginput of the opamp willbe higher than that at the non-inverting input only dur ing the positive half- periods of U.. This means that the drain-source channel of Ty Is switched on by the opamp when the voltage at the cathode tends to be- come lower than that at the anode. In that ease, the current through the FET flows from source (o drain, parallel to the in- ternal protection diode. In other words, the rer Is used the wrong way round. The forward voltage of the Ft diode s0 created is the product of the current Uarough it times the on resistance (0.07). The setting of preset P; determines the anode-cathode potential at which the out- ut voltage of the (mainly) linearly oper- ating opamp begins to rise and thus drive T} into conduction. The preset can be adjusted accurately only with the aid ofan oscilloscope connected to the drain and source of hte Fer. It is set to that position where for the nominal forward current the voltage across T; is as small as possible during the hall-periods when the Fer ison. In the prototype, the forward voltage so measured was 0.5 V with an alternating current of 10 Aata frequency of 50 Hz, At 3.3 A, the drop was only 0.2 V and at 300 mA just 0.1 V. Note that the forward voltage remains constant with currents below the level at which Py was set The circult draws a current that is not much higher than the supply eur rent toICy. Although the maximum supply voltage of the opamp Is 36 V, the cath ode-anode voltage, that is, the ‘reverse voltage ofthe “diode” must not exceed 20, which is the maximum permissible gate- source voltage of the BUZ10. (B. Zschocke - 924098) DECIBEL STEPPER Nhe stepper presented here makes use of nine positions of a 12-position rolary switch. It is, of course, possible to add the omitted steps. As usual, the amplification factors are arrived at by connecting a number of resistors in series in the feedback circuit. The re- sistors used here are from the E-96 se- ries, which enable a fairly accurate ap- proximation of the wanted amplification. I greater precision is required, combl- nations of two resistors may be used. This has the drawback, however, that a make-before-break switch must be used to prevent the output constantly being, switched to the supply line, ‘The bandwidth of the amplifier is de- termined by the set gain and the gain. bandwidth product of the opamp. If an LP351 is used as shawn, the gain-band- width product 1s 4 MHz, while the slew rate is 13 V ps“! ‘The eircult draws a current not ex ceeding 2 ma. Capacitor C; improves the stability when the amplification factor is large. (Amrit Bir Tiwana ~ 924065) sous 41 BLPKTOR ELECTRONICS DECEMBER 199280C552 MICROPROCESSOR SYSTEM ‘ere is something for microcontroller ‘enthusiasts to grind their teeth on. The 80C552 microcontroller from Philips Components is an upgraded derivative of Intel's 8032. It costs little more than the 8032, yet offers the following extras: (1) eight analogue inputs; (2) a 10-bit A-D converter; (3) a Timer?" with many extra features: (4) a ‘Timer3" watchdog Function; (5) an on-board 12C interface [6] 16 1/0 lines: (7) two pulsewidth mod. tulation outputs; and (8) a 16-MEz clock. Here, the 80C552 is used ina single- board microcontroller application, which 4s intended as an experimental system rather than a replacement for an exist. ing processor. The board is aimed at ver- salility. and accepts EEPROMs, EPROMs, or RAMs, or a combination of these, as memory devices, ‘The 80C552 board offers a multitude 1@ “Teeee] Legse > ELERTOR ELECTRONICS DECEMBER 1992of 1/0 connections for your own appli cations. All 1/0 lines, except PO and P2, are accessiile via connectors Ky. Ka: Ky and Ks. The [2C lines, Pg and Py.7, are taken to a simple 12C interface around a 6-way mini DIN connector. Kg. The TxD and RxD lines are available for TTL-level serial communication via two pinson con nector Ks, which also carries read, wnite, interrupt and limer signals. ‘Those of you familiar with the 8032 processor will find that the address de- ses coding cireult used here is fairly exten sive. Also, unusually, the reset input of the CPU 1s connected {0 a bistable. As to the address decoding, (his must meet a umber of special requirements: it must be possible to read instructions (using PSEN\) as well as read data (using RD\), and write data (using WR\). Further, it is possible to swap the address ranges ofthe ICyand ICs positions, with the CS1\ [chip enable) inputs of the respective EPROMs (or EEPROMs} used tosselect and de-select them, so that the de-selected EPROM hardly consumes power. When the processor is switched (0 ‘idle’ mode, ils current consumption is only one third of the normal value, while the CS1\ inputs of both EPROMsare automatically taken high. thus reducing power con: sumption even further: Bistable IC), and XOR gate IC, Se (o swap the positions, 0000H-IFFFH PARTS LIST Rosistor 1 ;R2,RS;R6:AB = 1000 | R7=B-way 10k0 SIL sC3:04;06 =100nF 10)F 18V radial (C8 =4uF7 16V radial ‘Semiconductor | 01 =1Naraa 2/D3,D4 = 56V 0.4W zener dlode D5 = N4OO1 ior =7aHoT74 iC2=74HCTe6 1C3.=74HCTOO 164105 = 27064 (EPROM) or 28(C)64 (FAM) or 64(C)64 (EEROM) ios = 74HoTS73 167 = PCBBOGS52-4WP (16-MH2 65- PLCC) \c8 = 7605, Miscellaneous: K11K2iKA;KS = 10-way box header. K9 = 14-way box neader. K6 = 6-way PCB-mount mini DIN socket X17 = 16MHz quartz crystal 68-way PLCC socket PCB Type 928071 ELERTOR ELECTRONICS DECEMBER 1992and 2000H-3FFFH, of the two EPROMs intheaddress map. Provided EEPROMS are used, this allows an interesting programming trick: reload the ‘uppet EEPROM (i.c., the one with the high est address) with the aid of a program. run from the ‘lower’ EEPROM. Look, no hands! No opening of cases, no ex. tracting of EPROMs, and no more time wasted on erasing and reprogramming ROMS. For instance, the program the ‘lower’ EEPROM may read the data supplied by the I2C or the serial input. organize it, and store it into the “upper EEPROM. Next, the program causes the ‘TO line to be pulled low and also stops triggering the watchdog. After some time, the watchdog will force an internal reset, which also pulls RST (pin 15) high for three clock cycles. This works as a clock for bistable 1C,,. The bistable copies the level on TO to its Q output. This enables ICzy, to swap the EEPROM address ranges by invertingaddress line 13, so that the program just loaded into the ‘upper’ EEPROM is executed. Because interrupt vectors are always Iocated from address 0000H, address ranges have to be swapped physically rather than in software. However. this ‘swapping is not allowed while the CPU isabout its normal business of fetching and executing codes, because an ad dress might change in the middle of an opcode fetch action. Ata clock of 6 MHz, this can be done with impunity, but def- initely not at a clock of 12 MHz or 16 MHz as used here, which forces us to use the hardware reset’ trick It should be noted that the analogue port, Ps, may function as an input only {it may also be used to accept digital lev els). The two °C pins, SCLand SDA (P15 and P, 7}, may be used as an input or an, output. Contrary to the other [/0 pins, they do not have internal pull-up re~ sistors, Further, it is recommended not to use P37 and Py g, as this will interfere with the normal opcode fetch operations. If you intend to connect an LCD (liq uiderystal display} module to the 80C552 system, itis best to use the LCD in 4-bit mode, because that allows the display to be connected to 7 port lines only A direct bus connection is not possi ble at clock speeds higher than 10 MHz (is. Walraven ~ 924071)gy SPEECH/SOUND MEMORY K's Series MSM6372-6375 enable _Resistor R, and capacitor C, provide output filter, provides an improvement Ihe reproduction of speech or other a power-up reset of the signal-to-noise ratio. sound stored in their internal ROM. Capacitor C,, which forms part of the A second channel enables the IC to The capacity of the ROM lies between 128 Kbit and 1024 Kbit. depending on [7 the type. The type and the sampling frequency specified [4 kHz. 6.4 kHz or SkHz) determine the length of the stored speech, which 1s 4-64 seconds. This time may be divided into 111 words that can be addressed individually Each IC contains a 12-bit digital-to- analogue converter (psc) and a fourth. order low-pass filter. The customer ean specify the words that must be stored on the mask-programmable ROM. As an example. the Type MSM6374-007 is, programmed to tell the time in English. Based on this IC. the circuit shown In Fig. 1 is intended to be connected to the FC 1/0 card published earlier this year of which the two converters have vera been omitted. The card then serves as interface between the FC bus and IC; The wanted word is selected via in. puts 10-16, Since the MSM6374-007 contains only words at addresses where 13 is zero, the corresponding input is strapped to earth. This arrangement eaves two of the eight available 1/0 bits for starting and timing of the words, These bits are available at sr(start input) and Nak (next address request output) Briefly, the control is walt until Nak is high: (er = raneri23 + key in the address; * wait not less than 10 seconds; roavose + briefly render st low (pulse duration 0.35-350 4s}. Normally the Nak signal indicates that the next address may be keyed in before the entire word has been spo: ken, This arrangement provides. smooth, transition between words or parts of To simplify the control of the syn: thesizer. the start pulse is not gener ated by software, but by monoflop IC, This stage is triggered by both the first and last transition of the start start sig. nal, which is applied to the trigger in. puts Via differentiating networks R,-C and R-Cy. The timing diagram in Fig. 2 shows what happens. Every time the computer writes data to IC,, the software inverts bit P6, It then takes 30 4s be fore the onset of the start signal, which, fs 250 us long. In this way, the control computer can start the ‘utterance’ of a word in one write operation. Without IC, the computer would have to writethead: dress first. then, after 10 4s, the start signal, and finally end the start signal. Resistors Rj. Ry, and capacitor C; set the oscillator frequency to 64 kHz. resulting in a sampling frequency of 6.4 kHe. ELEKTOR ELECTRONICS DECEMBER 1992ve speech with echo, two tones, or a tone with three different volumes. This facility cannot be used in this applica tion owing to lack of 1/0 bits. Input 2CH is, therefore, disabled by strap- ping it to the positive supply line. ‘The output signal is raised to about 1.W into 8 © by integrated bridge am plier 1C “The circuit may be controlled via the camputer-to-EC interface by software thal. once installed, may be enabled by a suitable key combination. It is possi- ble for either the time to be spoken or an alarm time to be set, The use of this software presupposes that the ’C driver (Type 1671—see p.110) has been in- stalled. When Program 177. is run, it installs itself, aller which it is enabled {even if another program ts being run) by the simultaneous pressing of keys eri and F1. If that combination ean. not be used, for instance, because itis used by another program. another com bination may be chosen by loading the program with tausmme/st. You will then be asked to key in the alternative com bination, Note that in the combination the program ean recogni hand shift key. When the combination has been keyed. in, a self-evident menu appears after about « second, pravided the screen is not in the graphics mode. In that case, a high tone is emitted, whereupon only functions ‘tell Lime’ and ‘alarm on olf are available, ‘Thecircuit drawsa current ofnat more than 800 mA. only the leit (OKI application ~ 924012) PARTS LIST Resistors: Bt =array, Bx100 ko R= 150 ko RB=47 ko. Ra, RE= 10k. RS = 100 ko AT = 22k Ra = 990 ka PL= TOKO preset ‘Capacitors: G1, ©5, CB, C10, C11 = 100 nF (02 =47 uF, 25 V, radial (03 = 100 pF C4=1 uF 25 V, radial C7, CB=1 nF 69-56 nF (C12 = 220 pF; 16 V, radial ‘Semiconductors: 11 =74HCT123 loz = MSMes74-007 IC3 = TDA7052 ‘Miscellaneous: 1 = 16-way box header, right-angled U4 =100 pH LS1 = loudspeaker, 8 0, 1 W Sotware Type 1771 (see p. 110) VOLTAGE INVERTER Ame enienly wltnge may be ob- lained easily with the use of 2 spe al inverter IC, Such ICs are not always readily available, however. Fortunately, the current drawn Is not excessive, standard HC-MOS chip may beused. The present circuit is based ona Type 74HC14, which contains six Schmitt triggers, whose combined gates can deliver a fairly high output current. Moreaver, such devices ran be made (o oscillate easily Parallel switching is, however, s prob- Jem with Schmitt triggers, even if they are contained in the same chip. Thi problem Is particularly acute in the ease of slow input signals. In the present cit- cuit this problem is resolved by driving, the parallel-switched input viaa gate that is not part of the parallel circuits. Since this is also a Schmitt trigger, the output signal has steep (fast) transitions, ‘The parallel-switched gates together with IC), form a reetangular-wave gen- erator, whose output frequency is about 125 kHz, The output signal ts converted bya charge pump into a negative supply voltage. The diodes used for that pur pose are Schottky types which, owing to their low threshold voltage, do not lower the load voltage by as much as silicon Biodes, ELEKTOR ELECTRONICS DECEMBER 1992 ary © ler =7aHera rev Under no-load conditions, the output voltage is about 6 V and the IC draws a quiescent current of around 100 1A. When the load current is about 1 mA. the output voltage drops to 4 V. If this voltage can drop even further (down (© half the supply voltage), a load current of up (a 10 mA ts possible. A higher lnad current or a smaller drop In the output voltage cannot be obtained by raising the value of the capacitors in the charge pump. since the IC cannot cope with this. Note also that the circuit is not short-circuit proof: the IC will not give up the ghost Immediately. but it does not take all that long. (. Pipers - 924081)60 WATT MUSIC AMPLIFIER [iit igarcbust no rtermedium power nplifier that is particularly suited 0 use in ‘combo’ type portable ampli- flers used by guitar players and jazz mm sicians. The amplifier isa straightforward combination of an integrated audio driver IC, the LM391-80, anda push-pull power output Stage designed with bipolar tran- sistors, A few peculiarities of the design will be discussed. The NTC, which is in ther- mal contact with the power output tran- sistors, enables the LM391 to switch off the power stage when this gets too hot The onset point of this thermal protec- tion lies at an NTC current of about 200 LA. The electrolytic capacitor shunt ing the NTC serves to provide a ‘soft start’that is, to prevent a loud click or other disconcerting noise from the loud- speaker when the amplifier Is switched on. It may happen that the protection is too sensitive, in which case some exper- imenting with the value of Ry, or that of the NTC, should be tried It is possible to implement feedback {n the amplifier by connecting Ryy to se ries network C5-Ry. The latter parts, to: gether with Ro, determine the frequency, response of the amplifier, which may need adjusting to meet individual re- quirements. The component values given, here will, however, be all right for most applications, The effect of different values of C5 and R; is simple to measure (or hear) by shorting out Ras temporarily. For 4-2 loudspeakers, Rag must be lowered to 0.18 Q. Unfortunately, the LM391-80 is prone to oscillation, which is suppressed by components Ry. Cg, Cs and Cg (in most eases, Cg may be omitied). Resistor Ry in particular reduces the open-loop gain. If Ry is used, Ry must be fitted to compensate the resulting off-set volt age. Components Ry and Cy form a Boucherot network that serves to stabi lize the amplifier at high frequencies. ‘The input of the amplifier should be driven by a low-impedance source capa ble of supplying ‘line’ level audio signals (0 dB). Network Rj-C, attenuates sig nals above 50 kHz or so. ‘The quiescent current of the ampli- fier is set by preset P). Set this control to 0initially, and adjust it until a qui- PARTS LIST A1g;R20 = osis/5W a1 = 101W R22 = 100/1W R23 Ax = 1MO (see tex!) Ry = 909k82 (see text) P1 = 1040 preset H Capacitors C1 = 2ur2 63v C2 =3nF3 not fitted (see text) C7;612;614;016 = 100nF C10;611 = nF C1;C15 = 10uF 63v ‘Semiconductors: Di;D2 = 1N4006 T1=B0297 72 =BD238 73 = BD250C T4= 8D249C ICt =LM391-80 Miscellaneous: Li=see text NTC =40k02 stud type Heatsink 1 KW ELERTOR ELECTRONICS DECEMBER 1992escent current of 50 mA flows, This may be increased to 400 mA if you are after low distortion. ‘The power transistors are all located at the same side of the PCB so thal they can be bolted on to a common heatsink, together with the NTC. The heat sink should be fairly large and have a ther mal resistance of 1 K W-1 or smaller Note that L; consists of 20 turns of 0.8mmdia. enamelled copper wirewound around Rp}. Ca is a ceramic eapacitor. Finally. some measured data (supply voltage: +35 V: Ry short-circnited); + 3-dB bandwidth (8.9): approx. 11 Hz to 20 kHz + THD (transientharmonie distortion) at 1 kez: LW into 8 Q: 0,006% (fy ~ 400mA) 1 W into 8 Q: 0.02% (ly = 50 mA) 65 W into 8 2: 0.02% (Li, = 873 mV) 80 W into 4 0: (Uy = 700 mv: onset level of current limit) (W. Teder et 1G 4h oN e IMPROVED LM317 REGULATOR Athos the properties af the M317 regulator are excellent, they can be improved by cascad: ing two of these devices. There is then a constant difference be tween the output voltages of the two regulators and, consequently aconsiant voltage across the input and output of IC: This arrange ment resultsin anappreciable im. provement ofthe regulating char- acteristics of IC, Moreover, its dis sipation is reduced so that the sta- bility of the output voltage with temperature is improved. Other properties, such as the maximum, output current of 1.5 A, do net change, of course. The output voliage, Uy, of the circuit depends on the ratio Re:P voltage drop, U4, across IC2 de pends on the operating point of IC} and may be calculated from: u 25(1+R2/Ri) {VI With values of these components as shown, Ly=3.5 V. It should be noted that this voltage must not drop below 3.0 V. Moreover, the value of Ry must be about twice that of Rj, and the minimum drop across the entire circuit must not be lower than Uiy#3 V. The circuit is highly suitable for use as a 5-V power supply. Iti, however, important that the di rect voltage al the Input is not lower than 12 V-This means that as follows: With values of these components as shown, the output voltage may be var ed over the range 1.25-11.5 V. The us 25(1+Rs/P)) IVI ELEKTOR ELECTRONICS DECEMBER 1992 the secondary voltage of the mains transformer must be 12 V instead of the usual 9 V. 924103)CHARGING TEMPERATURE MONITOR Hf monitor is particularly intended as an aid in the rapid charging of NiCd batteries. Most commercially avall able fast chargers have no temperature sensor, although temperature is an im. portant factor in the fast charging of NiCd batteries. According tomanufacturers datasheets, temperatures for various states ofa NiCd battery should roughly be: (a) 30°C when hhalf charged (b) 37°C when fully charged: and (c) 48°C when 20% over-charged. The monitor is connected in series with the battery and the charger. The tem. perature is assessed by a couple of NTCs (resistors with a negative temperature co" eflicient). The output of these devices is ‘compared with reference levelsin ICyyIC ‘The output of comparator IC 4 goes low when the temperature—measured by Rig and Ris—rises above 25 °C, where. upon Dy lights. The output of IC jc goes Iow when the temperature reaches 38°C, whereupon Ds lights. This signals that the battery is fully charged. Finally, the output of ICyy, goes low when the (em- perature reaches 45 °C, whereupon Dg lights. This 1s an alarm signal calling at- lention to the battery beingover charged. AL the same time, T; is switched on whereupon relay Re; is energized and sts contaet disconnects the charging cur rent. The charging current can be re connected only after the temperature has dropped (o well below 45 °C. Diodes D)-Dg ensure that only one of the three LEDs can light at any one time, The circuit is calibrated with P; to make certain that the LEDs light at the cor- reet temperatures. When the monitor is connected to the charger, itis essential that the charging current, originally flowing via Ky, flows via Ky and Ky. The batteries must, there: fore, be connected to Ka and Ky. The ter minals of the NTCs are also coupled to these connectors. ‘The NICs must be coupled to the bat teries in a manner that ensures good contact. It is good practice to fit them permanently to the batteries. ‘Thesupply forthe monitormaybetaken from the charger, either via the 12 V pin or the L pin. In the latter ease, wire link JP; must be short-circuited. The moni- tor draws a current of about 15 mAwhen the relay is not energized. Variations in the supply voltage do not affect the monitor since the NTCs are connected in a bridge arrangement. seit & ot-4 s et} 4PARTS LIST id= 1NAO0T 1 = BC557B Capacitors: 11 = M338 C1, C4 = 100 nF C2=6.8nF Miscellaneous: 3 =470 nF Kt =S-way terminal block, pitch 5 mm Semiconductors: Ke =5-pin DIN socket D1-D6,D10=1N4148——-K3=5:pin DFIN plug D7=LED,3mm, green Ret = 12 V relay for PCB. DB =LED. 3mm, yellow mounting D9 =LED, 3mm, red As shown, the monitor is intended for use with 12 V batteries, It can be modi- fied for use with 6 V batteries merely by using a 6 V relay instead of a 12 V type (C. Millasson ~ 924066) DIGITAL PATTERN GENERATOR Dininethsconsiuctionand testing digital cireuits, there is often a need ofan accurately defined bit pattern. The generatordeseribed here uses an EPROM tostore the bit information for the desired pattern, | When switch) is pressed. a given pat- tern is generated once: seven different patterns can be generated simultane- ously. The eighth data output of the EPROM is used to mark the end of a pattern. After S; has been pressed, gate ICi arranges the resetting ofaddress counter ICp. Gate IC ¢. which is connected to the input of the address counter. functions asa start stop oscillator that is switched bby two NAND gates. These gates are de- signed to keep the oscillatordisabled until the switch is released, Only then are pulses fed to the address counter. The fr: quency range of the elock can be varied toindividual requirements by altering the value of C3. With values as shown, the frequency canbe set with P, between 15Hz and 150 fz. When it patterns are being generated, data line D7 must be logic high. At the end of the pattern. this line goes low whereupon the clock is stopped. ‘The maximum length of a bit pattern, in the circuit as shown is 8191 clock pulses (8x1024-1 stop pulse). Iflink JP, fs set to position B, an external stop pulse can be used, ‘The circuit draws a current of about Sma, (G. Kleine ~ 924033) } | : ELEKTOR ELECTRONICS DECEMBER 1992PLL SYNTHESIZER FOR TV RECEIVERS Meee toners normally contain means to scale down the VCO (volt age-controlled oscillator) signal. The one used in the proposed synthesizer is a Philips Type UV816/6456, whose seale factor can be set to 64 or 256. It oper- ates over the low VHF band, the high VHF and hyperbands, and the UHF band. ‘The proposed circutt Is based on a Siemens Type SDA3002 frequency s| thesizer IC that may be controlled by a computer. It works exclusively with a .64 scaler. This is ensured by leaving pin 15 of the tuner open. Network Ro-Ca-C) forms the filter for the phase-locked loop—PLL. The level of the charging current for the filter 1s de- termined by the value of R; and bit 14— see the table. In the prototype. both low and high levels of current gave stable PLLoperation, but the loop reacted faster with a high current. ‘The SDA3002 has a unique oscillator for providing a reference frequency. The frequency of erystal X; is scaled down internally by 4096, so that the reference frequency for the PLL ts 976.5625 Hz. ‘The PLL is set by enteringa data word into the SDA3002, for which a clock. °PL, an enable signal. PLE, and a data signal. IFO. are needed. At each trailing transition (edge) of the clock, a bit Is shified into the IC, provided PLE ts high— see Fig, 3. Only when PLE has become low will the PLL accept the entered data into a latch, A total of 18 bits must be shifted into the chip, Fourteen of these contain the frequency setting of the PLL, The 15! bit determines the charging current. When this bit is high, the current is 10/p, where /; is the reference current; when it is low, the charging current is f. The 16% bit controls the NORM output. The final two bits determine the state of the band selector outputs: pins 4-7, see the lable. ‘The first 14 bits are computed fairly casily once the desired channel frequency, Jo. the intermediate frequency. if (38.9 MHz), the scale factor, 2 (=64) of the tuner: fy. of the PLL are known. The overall ‘scale factor, 2, is then aU [I/ 20h The result should be rounded to the nearest whole number, and then split into two parts (IC) contains a dual-mode prescaler) (0 obtain the bits needed by IC). The nine most significant bits— MSBs—are calculated by dividing z by 32 and ignoring the digits following the decimal point. The remainder of 2:32 forms the five least significant bits— LSBs. As an example, assume that channel 29is wanted; the carrier frequency is then 535.250 MHz, The scale factor is 135.25+38.9)/64x976.5625= =9186, Dividing that number by 32 gives 287, remainder 2. In binary form, that is (MSB)1000111 1100010 (LSB). However, the data must be entered in inverted form, thai is, a logic 1 corresponds to low and a logic 0 to high. Taking the LSBs first, the following levels must ap- pear at the data input HLHHHLLLLLHHBL. Diode Dj. which is controlled by the LOCK output via T}, then lights to indi cate that the PLL is locked, ‘The maximum tuning voltage is de termined by the supply voltage, which is connected to the loop filter via Ry, The UV816needsa tuning voltage of 0.7-28V. The SDA3002 can handle a maximum voltage of 38 V at its UD output. 1 oc yyy Video Demodiator IFO bit2!4 | pump current i ir H 10%, IFO bit 215 | NORM ouput L L W H TRO bit band selection output 216 217 Le LH Fy HOW CTRONICS DECEMBER 1992‘Two further supplies are needed: 12.V for the tuner and 5 V for IC; and the prescalerin the tuner. Since the combined current drain of IC and Dj (lighted) is only 29 mA, the 5 V rail is easily ob tained from the 12 V supply via 5 V reg- ulator. That regulator can at the same ime provide the current—about 25 mA— for the prescaler in the tuner. Additionally the 12 V supply must provide the cur rent for the tuner, which in case of the UV816 is about 85 mA. A total current of some 140 mA is, therefore, required. (7. Giesberts - 924072) 3 lh em || E Db ous ff omem }] A z E pA ESAT Hey aml oeer | ve ce ean INDUCTIVE PROXIMITY SWITCH nductive proximity switches are used, for instance, for measuring motor speeds or determining the position of metal ob. jects, They do not suffer from mechani- cal wear or sparking contacts . The lat- S wee ofv HN) : i ELEKTOR ELECTRONICS DECEMBER 1992 ter ts particularly important in spaces where explosive materials are stored Virtually all commercial proximity switches are constructed as shown in Fig. 1. An inductor in the resonant cir cuit of an oscillator serves as the sen- sor. If a conducting object enters the magnetic field of the coil, eddy currents areset upin the inductor, Thisdamps the resonant circuit so that thevoltageacross drops. This voltage drop is monitored with a Schmitt trigger. When the object gets very close to the inductor, and the voltage across the circuit drops suffi- ciently, the Schmitt trigger changes state. ‘The trigger is followed by an output stage. ‘The sensor. IC}, used in the present circuit translates the approach of an ob- Ject intoa fallingcurrent through the sen- sor. In the absence of an object, the eur- rent is about 4 mA; when an object is at a distance of 4 mm from the sensor, the current is only 1 mA, ‘The sensor current is converted into a voltage by Ry. This voltage is applied to the non-inverting input of Schmitt trigger IC2, Actually, since the hystere- sis is small, IC» functions more as a ‘comparator that likens the voltage across R, to that across Ry. When an object is within the proximity limit of 5 mm of the sensor, the potential across Ry is larger than that acrass Ry, whereupon the out put of IC) becomes logic high. ‘The potential across Ry is dependent on the supply voltage, of course, but over the range of supply voltages stated §n the diagram, it is always greater than the smallest drop across Ry and always smaller then the largest drop across Rj, This ensures correct switching of [C2 in all situations. The only quantity affected by the supply voltage is the output potential of the eireutt. ‘The sensor is a Type IFR10-82-01 from Baumer Electric. It has a diameter of 10 mm and is 5 mm long. UUs. Mossbauer ~ 924073)SERIAL DATA GENERATOR 1e data generator offers a simple means of testing a cireuit.There is no need of software for some micropro. cessor and all the components are read. ily available. The design is based on the timing of a Type SDAS002 elreuit. In contrast to the two-wire 'C format, a three-wire connection is used. If only one fixed data word needs to be sent, DIP switches S, and S, and the eight fold pull-down resistor arrays, may be omitted: the inputs of IC, and IC, are then held at fixed levels, ‘Thegeneratorisbased on two cascaded shit registers, IC, and IC,, that are used as 16-bit parallel-to-serial converters. The clock is provided by IC,, which is started by setting bistable IC,,, with press button switch S,. When this switch is pressed, a 100 j1s pulse is applied via differentiating network Rj-C, to the set input of the bistable, The capacitor is discharged via R,. The pulse duration is not changed by keeping, depressed: this " is necessary because if it were to exce 3080 TancTee Taos BLEKTOR ELECTRONICS DECEMBER 1992a half clock period, the timing would go astray. The output of IC), is the ENABLE signal for the circuit to be tested. The start pulse is used also to load the logic levels set with the DIP switches on to shift registers. The registers ac- cept the parallel data as long as the sig- nal at their pin 1 is low. This is another reason that the start pulse has to be short The start pulse begins the entire test cycle—see the timing diagram in Fig. 1 Output 25 (Q3, pin 7) of IC, 1s inverted by IC, to provide the clock for the shaft registers. This ensures that DO is pre- sent al the output for an entire clock period (1.25 ms). Alter output 2° of IC; has been high 18 limes, the bistable is reset and the enable line is pulled low by IC). which is connected to outputs 2! (QS. pin 4) and 2¥(Q8, pin 13) of C,. This completes the eycle. In this example, 18 bits are shifted. ‘The two MSBs are determined by the se ial input (pin 10) of IC. If this input is high, D16 and D17 are ‘I’, If these bits are required to be variable, one or (wo additional D-type bistables, connected as shift registers, must be added ‘The current drawn by the generator. determined primarily by the pull-down resistors, is very low: if all bits are high. ilisabout 801A. After, has been pressed, the oscillator ts started briefly, which dow: bles the current, (P. Giesberts - 924076) TEMPERATURE-FREQUENCY CONVERTER Tis ctreuit converts temperature jeasured by an NTC (negative tem: perature co-efficient) resistor into a dig- ital signal. The resistance of the NTC which is an inverse function of the am. bient temperature, determines the fre quency of an oscillator built around the familiar TLCS55 timer. The astable cir cuit is designed to give a pulse repeti- tion frequency (p.r.£} of about 250 Hz at 25 °C that rises with temperature. The non-linear relationship between tem: perature and oscillator frequency is not a problem here, because it is relatively simple to ‘straighten’ by arithmetic op: erations performed by a computer or a microcontroller system. Basically, three temperaturesare mea: sured, and the corresponding oscillator frequencies stored as reference points, which serve to interpolate other values in between, The converter is built from SMA (sur- face-mountassembly) componentson the printed circuit board shown here. It is, however, possible to use a standard NTC {in position R, To ensure fast response to tempera ture changes. the completed PCB is fit- {ed into a small metal tube with a dsam. eter of 13 mmor larger. Greateare should betaken to isolate the board and the corn: ponents from the metal tube; if neces- sary, use heat-shrink sleeving! The tube fs sealed hermetically with potting com: pound or two-component resin glue, through which three wires are passed: + supply: output signal: ground, Current consumption ofthe converter 4s smaller than 1 mA, ‘The multifunction measurement card for PCs (Ref. 1) is perfect for processing the converter output signal. Use is made of the frequency meter input and the program modules found in the Turbo Pascal library PMEASURE. PAS’ on diskette ELEKTOR ELECTRONICS DECEMBER 1992 ESS1751—see p. 110, and in Borland’s ‘Numerical Toolbox’, which provides some handy interpolation routines, (U, Kunz ~ 928020) Reference: “Multifunction measurement card for Pes", Elektor Electronics January and February 1991 le =TLesss00 PARTS LIST {all parts surface-mount assembly) Resistors: 47kQ. NTC ‘Semiconductors: D1;02= BAS32 IC1=TLCS55CDMAINS-POWER-ON DELAY his circuit provides a simple means of implementing a ‘soft-start’ on heavy mains loads. Basically, this is achfeved with the aid of high-power se ries resistors and a relay. The eircui also enables the load to be switched on with a small (light-duty) switch ‘Two relays are used to connect ar by. pass a series of power resistors inserted between the mains'live' wire and the load. ‘The ‘solt start’ is achieved by first con- necting a resistor between the mains and the load, and then short-circulting the resistor again so that the full mains voltage is applied to the load (equip: ment). The purpose of ‘soft’ switching is toprevent fuses blowing asa result of very high surge currents, The relay coils are connected in se- ries: the reactance of capacitors C, and Ca keeps the current through them at about 27 mA. The direct voltage across thecoilsis 48V, When the delay is switched on with S;, relay Rez comes on well after Re} because its coil is shunted by two large electrolytic capacitors, which take some time to be charged from the cur- ent source formed by the mains and capacitors Cy and Ca, When Reg is ener ed, its contact short-circuits the power stors. The simplicity of the eircutt gives rise (oaminor disadvantage: some time must be allowed between switching off and switching on again to enable Cy and Cs to discharge. Depending on the length ofthe switch-on delay required, the values of Cy and Cs may be changed a little. It should be noted, however, that the surge current is limited toabout 5A fatamains voltage of 240 V) by Rs-Rg, and that the resulting dissipation equals about 1200 W in four 5-watt resistors! Therefore, wher alonger switch-on delay is wanted, higher wattage resistors (the PCB can accom: modate fairly large types. which may be fitted vertically) must he used. ‘The current drain ofthe circuit is prac tically that of the two relay coils 1n se- ries, that is, about 27 mA. The current through the neon lamp is negligible. The specified relays have contacts rated at 16A. Ifthereis reason to fear that the PCB mounted terminal blocks and PARTS LIST | Resistors: Ai =2700 R= 1MQ. FG-Re = 100 5W Capacitors: C1 = see text C2=see text | C3 = 22uF 40V radial (C4 = 1000uF 40V radial C5 =470uF 40V radial ‘Semiconductors: B1 = B380C1500 (380V piv, 1.5A) Miscellaneous: K1:K2)K3 = 2-way PCB terminal block, pitch 7.5mm. S1 = on/off; mains-ated 1A. Ret;Re2 = V23056-A0105-A101 (Siemens). cchassis-mount neon lamp with integral resistor. 4 spade (fast-on) screw-on type plug for PCB mounting, Lat o 00 0 a ELEKTOR ELECTRONICS DECEMBER 1992their solder connections are not up to the load current, parallel-connected spade terminals (fast-on’ type) should be used instead, Finally. the valuesofsome components depend on the mains voltage and fre- queney as follows: 220V, 50Hz: C1=330 nF: Cz: 230V, 50Hz: C, = 470 nF: Cy 240V, 5OHz: C; = 470 nF: Co 120 nF ‘not fitted not fitted MAINS POWER-ON DELAY (Capacitor working voltage: 680 VDC 7250 VAC) L1OV, 60Hz: C, = 680 nF: Co = 470 nF: also Ry = 120.2. (Capacitor working voltage: 400 VDC 7130 VAC) WARNING This circuit is connected directly to the mains, and carries potentially lethal volt ages. Never work on the circutt, or ad. just it, while itis connected to the mains. ‘Observeall precautions relevant towork- ing with equipment or components con- nected to the mains, (T. Gtesberts - 924055) ‘The order forms are on page 111 this month CAPACITANCE METER Of eben neta wae here is able to measure capacitances between 100 pF and 1 1 Fover five ranges. The circult consists of a variable os- cillator, a sealer and a measuring stage. The oscillator Is based on an inverter contained in a’ Type 74HC14 and gener- ates a frequency, f. that Is inversely pro- portional to the capacitance between terminals C,, Roughly. LRG, where R depends on the position of $) With values as shown, the frequency lies between 240 Hz (Cys uF) and 12 kHz (G=100 pF} Scaler ICg divides the output frequency of the oscillator in a manner which en: sures that the maximum output fre: quency in each range is 120 Hz, ‘The measuring stage is driven by cur- rent source T). During one half of each period of the output signal of IC; capac 1s charged via Ty. During the other half of the period Ty is switched on by the signal, so that Cz 1s short-cir- cuited. In this way, the maximum po: tential across C2 depends on the fre. quency of the signal. The potential is buffered by opamp ICy,, and integrated by R}-C). The resulting direct voltage is used to deflect the meter via opamp IC. ‘The circuit is calibrated by connect inga known capacitance of about 100 nF (0.1 4F) across the measurement termi- nals and adjusting P; so that the meter reading corresponds to the capacitance. ELEKTOR ELECTRONICS DECEMBER 1992 Since normally the meter 1s used only ‘occasionally. it may be powered simply by a 9.V PP3 (6F22) battery. [R. Shankar ~ 924063)FLASH EPROM CONVERTER ash-EPROMS are becoming increas ingly popular, in spite of their being harder to reprogram than BEPROMs. That difficulty is, however. countered by their lower price, greater density and higher programming speed. Maxim. a specialist manufacturer of all soris of small converter, produces a special IC for generating the necessary programming voltage of 12 V at 50 mA: the MAX732. ‘The MAX732 contains virtually every thing to make a mini switch-mode power supply, Its input voltage requirement Is 5 V, from which it produces an output of 12 V. Since that output is needed only during programming, it may be disabled via the shutdown input (pin 1), A problem encountered in producing switch-mode supplies is the availability of certain passive components, particu: larly the inductor. That used here is the Sumida Type CD54-470KC, which is available from a number of retailers and also from Maxim dealersas"Type MAX-L001 Ifmeither of these can be obtained, how- ever, a triac choke may be used, but that will lower the efficiency of the converter to some extent. It 1s always possible to add or remove turns if the inductance is Incorrect. Bear in mind that the indue- tance {s proportional to the square of the number of turns. Diode Dy must be a Type 1NS187 as indicated or equivalent; note that a IN4001 is not fast enough. ‘The prototype delivered 12 V and an output current, jy, of up 10200 mA, more than enough for a flash-EPROM. The current drawn from the 5 V supply was about 2.44, Use a single earthing point and de- couple the IC directly at its pins, (Maxim application - 924111) ‘The subscription rates are on page 112 this month SOLAR-CELL SUPPLY "Ti simplest solar cel supply stem consists of three parts: a diode. a solar-cell panel, and a rechargeable bat- tery. The diode prevents the battery being discharged through the solar panel when no or little sunlight falls on to the panel. Although the diode is usually a Schottky type, the forward drop of this type may ‘cause an appreciable loss of energy. Since the low-500 kHz. Aslong as the core material is not saturated, the sensitivity ofthe probe is about 0.2V AT (B. Danker - 924103) COMPACT RS232 ISOLATOR Tes mins from Maxim is a special IC for designing an electrically iso- lated RS232 connection. Such links are particularly important when several pieces of equipment (which may or may not be Isolated from the mains) are to be inter- connected even when they have differ ent earth potentials, The chip is available in two modes: A (expensive) and B (inexpensive). Note, however, that even inexpensive is rela. live, because the B type still costs mare than £30. The maximum isolation voltage that the B-type can handle for one minute is 500 V, whereas the A-type can handle up (o 1250 V. Moreover. the A-type can have a constant voltage across its ter- minals of 130 V. The maximum trans. mission speed of both types is 9600 baud, ‘The internal transformer of the chip provides the supply voltage for the right hand side of the diagram. The input and output levelsat the left-hand sideare TTL. compatible. The inputs have a hystere- of roughly 0.5 V and pull-up current sources of 0.4 WA for the suppression of noise signals. The inputs and outputs at the right hand sideare RS232 compatible. Outputs Tig, aNd T2e,, deliver a voltage of 7.5 V oa load of 3 ko. With a supply voltage of 5 V, the IC draws a current of 90 mA, Pin 11 of the MAX252 is an active low. output-enable inputand pin 3an ae- live-high shut-down input. A high level so] maxzs3 at pin 3 causes the 190 kz oscillator of __ S232 compatible levels at the left- the integrated supply to be switched olf, hand side canbe provided by the MAX: Tq and'T2,q tobecomehigh-impedance, as shown in the diagram. This IC also the-4 WA pull-up inputs to be deactu- arranges the necessary inversion of the ated, and the power acceptance to be various signals. reduced to 50 jW. (Mtaxim application ~ 924107) ELEKTOR ELECTRONICS JECEMBER 1992DUMMY LOADS ARTIFICIAL AERIALS FOR TRANSMITTER TESTING A dummy load is an artificial aerial that is used for making measurements and performing tests on radio transmitters without actually radi: ing a signal into the air. Radio operators should routinely use dummy loads to tune up on crowded channels, and only when the tuning is completed transfer power to the live antenna. In some countries, this procedure is not merely good manners, but required by law (even if often ignored). (OTHER use for dummy loads is in {troubleshooting antenna systems con- taining multiple elements (e.g. tuners, coaxial cable, low-pass filters. and so forth). Suppose, for example, we have an antenna system in which the VSWR (volt- age standing wave ratio) is high enough to adversely affect the operation of the radio transmitter. Modern transmitters, with solid-state final RF power amplifiers, have VSWR-sensitive power shutdown circuitry, We can test such a system by disconnecting each successive element in sequence, and connect the dummy load to its output. If the VSWR goes down to the normal range. when a particular element is replaced with the dummy load, the dif ficulty is probably distal 10 the point where the dummy load was inserted (i.e. towards the antenna). You will eventually find the bad element (which is usually the antenna itself), Another use for dummy loads is in test- ing for television, broadcast radio or audio system interference. Once itis established that it is your transmitter that is causing the problem, it is necessary to determine whether or not the route of transmission is. through the antenna, around the cabinet flanges or through the AC power mains connection. If the offending signal is from the antenna, then the root cause might be not the signal or transmitter, but improper filtering or shielding of the television or other appliance, or the inability to handle local overload conditions ereated by (but not the fault of) your transmitter. Ifthe in- lerference persists in the face of using an RE dummy load substituting for the aerial, then the problem is in the power cones tion or flanges ...and you must do som thing further to find and suppress. the faut Forms of dummy load A common, but irregular, form of dummy load is shown in Fig. 1. It consists of a 40, to 100-watt electric light bulb that is con- By Joseph J. Carr, KaIPV Fig. 1. Light bulb dummy load, nected {0 a length of 52-0 coaxial cable and a connector that mates with the trans- mitter. The centre conductor of the coax- cable is connected to the centre button Of the light bulb, while the shield of the coaxial cable is connected to the outer threads or base (depending on country of origin) of the bulb. Tt was once common practice to paint the bulb with either aluminium or copper conductive spray paint, except for a small window to see the light level, The paint supposedly shields the bulb and thereby prevents RF radiation, That supposition, however, is highly optimistic. One day about 30 years ago, I used this type of dummy load to test my Heathkit DX-60B 90-watt CW transmitter. A friend of mine answered my ‘call” (supposedly made to a dummy load), and reported an S7 signal strength...rom a distance of nearly ten kilometers! That is not exactly how a ‘dummy load is supposed to work. Another defect of the light bulb ELEKTOR ELECTRONICS DECEMBER 1992 i ee el hE lc =" le ly dummy load is that its resistance changes with light brightness. Hence the bulb is ‘not stable enough to be seriously consid- ered as a dummy load except in the crud- est sense, The light bulb dummy load, while cheap and easy to obtain, is too much of a problem for all but impromptu emergency Situations, It is nor recommended. Figure 2 shows the most elementary form of regular dummy load which con- sists of one or more resistors connected in series, parallel, and/or series-parallel as needed to make the total resistance equal (o the desired load impedance (usually 50). The power dissipation of the dummy oad in Fig. 2 is the sum of the individual power dissipations. By using ordinary 2- watt carbon composition resistors, it is possible to make reasonable dummy loads fo powers of about 50 watts. Above th power level, one must consider the effects of stray capacitance and inductance from all of the resistor leads and interconne: 1g wires. Higher levels can be accommo- dated, however, if care is taken 10 keep capacitances and inductances low. It is essential that non-inductive resis tors be used for this application. For this reason, carbon composition or metal film resistors are used. There are actually two forms of non-induetive resistors on the Fig. 2. Simple, low-power dummy load mac from carbon composition or metal film non- inductive resistors.1 AND MEASUREMENT a Fig. 3. Commercial 5-watt dummy load. = Fig. 4, Commercial 50-watt dummy load. Fig. 5. Commercial 50-watt, 50-ohm dummy load (courtesy Bird Electronics). market. The carbon composition and metal film types are intuitively obvious. The other form are wirewound resistors in which adjacent turns are wound in oppo- site directions so that their mutual mag- netic fields cancel each other out. These are called “counter-wound’ resistors. For very low frequency (< 20 kHz) work. itis, permissible to use such counter-wound wire low-inductance resistors. These re- sistors, however, cannot be used over @ few hundred kilohertz Several commercial dummy loads are shown in Figs. 3 through 10. The dummy load in Fig. 3 is a S-watt model, and is typically used in Citizen's Band and other low-power (QRP) HF transmitters. The resistor is mounted directly on a PL-259 coaxial connector. These loads typically work to about 300 MHz, although many are not really useful over about 150 MHz because of Stray capacitance and induc- tance. A higher power version of the same type is shown in Fig. 4. This device works from VLF to the low VHF region, and is able to dissipate up to 50 watts, T have used this dummy load for servicing high VHF landmobile rigs, VHF-FM marine and low-VHF landmobile rigs, ay well as ham radio rigs. The load resistor in Fig. 5 is a Bird Electronics Termaline load that works at power levels to 50 watts, while presenti 850-9 load, ‘A. 300-watt_ amateur radio S0-ohm dummy load is shown in Fig. 6. This one is made by MEF Enterprises in the USA. Unlike the other two models, above, it is built inside a sheet metal cabinet, and is low in cost ‘An MFJ Versafoad is shown in Fig. 7. This dummy load is similar in form to the old (now off the market) Heathkit Cantenna, It consists of a high-power, 50- Q non-inductive resistor element mounted inside a paint can that is filled with ordi- rary motor oil. The oil increases the dissi- pation capability of the resistor, but to seep out of the load if it iS not well sealed, Very high powered loads are shown in Figs. 8 and 9, These devices are Bird Electronics “coaxial resistors’, and oper- ate to power levels of several kilowatts or more. The high power load of Fig. 9 is cooled by flowing water through the body of the resistor, and then exhausting the heat in an air-cooled radiator. Our final dummy load resistor is shown in Fig. 10. The actual resistor, made by R.L. Drake Co. in the USA, is shown in Fig. 10a, while a schematic view is shown in Fig. 10b, The long, high-power non-in- ductive resistor element is rated at 50 Q. and can dissipate 1000 W for several min= utes. If longer times, or higher powers, are anticipated, then forved air cooling is ap- plied by adding « blower fan to one end of the cage (on the Drake product, a remov- able mounting plate for a 3.5-inch fan is provided), ends Providing an output level indicator The dummy load in Fig by the author by addi for RF signal sampli nected internally to either a two-turn loop made of 22 AWG (approx. 0.8 mm dia.) insulated hook-up wite, or a 25-cm brass rod that is positioned alongside the resis- tor element (as shown in Fig. 10b). The loop or rod will pick up a sample of the 10 was modified the BNC jack (J2) Fig. 6. Amateur radio dummy load for VLF to 180 MHz (courtesy MF. Enterprises, nc.) Fig. 7. Dummy load mounted inside a paint ‘can (Courtesy MFJ Enterprises, Inc.) Fig. 8. High power dummy loads (Courtesy ‘f Bird Electronics), ELEKTOR ELECTRONICS DECEMBER 1992Fig. 9. High power, water-cooled dummy load (courtesy Bird Electronics). so that it can be viewed on an oscil- loscope, or used for other instrumentation purposes. Figure 11 shows an oscillo- scope photograph of an amplitude modu- lated RF signal taken from my modified dummy load. The transmitter was a 60: watt AM rig modulated by a 400-Hz sine wave from a bench signal g Another approach to providing an out- put indicator is shown in Fig. 12a. In this, case, a germanium signal diode (N34 or INO0 is suitable) is connected to the end ‘of the signal sampling rod that connects to the output BNC jack. The diode rectifies the signal picked up by the rod, or sam. coil if one is used, and the R-C net 1 filters it to remove residual ‘A variation on the theme is shown in 12b. This circuit, which is like one that was used in the Heathkit Cantenna, uses a resistor voltage divider (R2-R3) connected across the dummy load (Ri). A germanium signal diode (IN34 or 1N60) is attached to the junction of the two volt age divider resistors. The voltage at the junction (U,) is re: lated 10 the RF output power applied to the dummy load by the voltage divider equation: V50xP where P is the RF power in watts, The Fig. 10. a) amateur radio dummy load, 50-ohm, 1000-watts Intermittent (constant when op tional fan is added); b) circult of dummy load, with author’s modifications. FLEKTOR ELECTRONICS DE MBER 1992 DUMMY LOADS Fig. 11. Typical AF display on an oscillo- ‘scope from pick-up unit inside dummy load. Fig, 12. 8) diode detector circuit for circuit of Fig, 10; b) resistor voltage divider for detect- ing RF signal output voltage from the circuit of Fig. 12b is quite reasonable. With the shown, a 100-watt transmitter si duces an RMS (root-mean-squat of the order of 16 V voltage Conclusion Dummy loads are used as artificial aerials that permit one to energize a radio trans. mitter for testing, troubleshoot, justment_ without ‘al. Their use is practice, is neighbours, and is leg ‘most countries. Use them, and we will all be better off :DELTA-PEAK NiCd CHARGER Design by L. Pijpers Fast charging high-capacity NiCd batteries with large currents does not only save time, but also prolongs their useful life. Moreover, if the charger operates on the delta-peak principle, the batteries are always charged timely and to full capacity. igh-capacity NiCa batteries from which Jorge currents are drawn benefit from being charged with large currents, Owing to the lange discharge currents free metal ions in the electrolyte are produced in the course of time. This raises the internal resistance of the battery so that it ean no longer provide such large currents. If the battery is charged with a large current, the ions are eliminated So that the internal resistance of the battery returnstoitsnormal value. Itshould beborne in mind, however, that not all NiCd batter iesean be charged with large currents, More ‘over, even if the battery is capable of being fast charged, this should never be attempted when itis not fully discharged, Finally, to ensure that the battery’ has a long lite itis recommended that every fifth fast charge is followed by a standard 14-hour charge. Delta-peak principle Jn standard practice, NiCd batteries must be charged at |, of their Ah capacity for 14 hours, which meansa slight overcharge. Fast charging requiresspecial measures. Asshown in Fig. 1, the cell voltage during charging, initially rises rapidly and then, alter a cer tain level has been reached only slowly. When the battery is almost fully charged, the cell voltagerises morerapidly again untilits fully charged, after which the voltage actually drops. This can be explained as follows. The charging current not only charges the bat tery, but also decomposes the electrolyte, whereby gases (particularly oxygen)are pro- duced that cannot combine with the elec trode material. The consequence is that the pressure in the (sealed) baitery rises slowly: When the battery isnearly fully charged, the charging current produces primarily gas and contributes little to the charge, This results in a rapid rise of the pressure and the cell voltage. Because of that pressure, part af the ‘oxygen combines with thematerialoftheneg- ative electrode which produces heat. Since NiCd batteries have a negative temperature coofficient of about 4 mV °C°, shortly after the battery has reached full charge, the cell voltage drops again. Figure 1 shows that the peak voltage is reached just after the battery is fully charged. The delta comes about because the prin. ciple used here depends on the monitoring ‘of small voltage changes, which are denoted mathematically by the Greck letter d{elta) With this principle, as soon as the cell volt age shows signs of dropping, chargingis dis- continued. The circuit “The prime function of the circuit in Fig. 2is the monitoring of small changes in the bat- tery voltage. To this end, the battery voltage isapplied toan integrator, consistingof R,and , sia protection diode D, and Ry The inte gator has a time constant, t, fs, As long, as the battery voltage rises, the potential ‘across C; remains slightly below the voltage at junction R-R). When the battery voltage drops, the potential across C:, owing to the delay, becomes higher than the voltage at junction R-R,, This comes about because of thechargingcurrentofC, which flows through Rand thus causes a voltage drop across this resistor. The small drop across Ry is applied to IC, (a FET that has very high impedan inputs). When the voltage at pin 3 of this ‘opamp is higher than that at pin 2, the out- put of the amplifier i high, but T; does not Yet conduct because T, is off since its base emitter voltage is held at 0 V via the relay contactand R-R,. When S; is pressed, T; con- ducts, T; switches on and the relay is ener gized, after which T, is held on by Ry the hat process has commenced. When the peak cell voltageis reached, the potential at pin 3 of IC; drops below that at pin 2. This results in the output of IC, going ow, whereupon T; is switched off. The relay is then deenergized so that Tis off charging has ended ‘Thiswould be teuein: real world opampshaveofiset voltagesattheir inputs so that they don’t switch at exactly OV. The offset voltage of IC is ‘nullified’ by setting P; so that the IC changes state when the voltage at pin 2 is about 2 mV higher than that at pin 3. The printed-circuit board in Fig. 3 has two terminals (off’) where the offset voltage can be measured with a digi- tal voltmeter The charger operates either from a 12 V (cae) battery or 2 12-14 V mains adaptor that ‘can deliver the necessary high currents. niddeal world: inthe Construction and setting up The charger is intencled to be constructed on the PCB shown in Fig. 3, All components bar the relay and P, must be fitted upright. Ibis, therefore, necessary to fit the seven salder- ing pins before the components, Power te- sistors R, and Rj are not fitted on the board. ‘The board should preferably be mounted in a small metal case to which Ry and R,: may be fixed for good heat condeution, During setting up,a discharged NiCd bat- tery must be connected to K, (but it must not Sik Belg” 3 ad ie 2” s g (CHARGE CONDITION (%] Fig. 1. Charge condition vs voltage/pressure/temperature characteristics, ELEKTOR ELECTRONICS DECEMBER 1992be charged, which itis notaslong as isnot pressed). Since the maximum current drawn Isthen 500 mA, the charger may be powered froma simple 12V mains adaptor or battery Link JP, must be set and a DV! lowest voltage range, connected to ‘otf’. Set set to its P, for a reading of -2.0 mY. Remove the link and the DVM, whereupon the chargerisready for use Asa refinement, a suitable 5 A ammeter for monitoring the charging current may be fitted permanently in the +12 V supply line Fig. 2. Circuit diagram of the delta-peak NiCd battery charg Fig. 3. Printed-circuit board for the delta- peak NiCd battery charger. ELEKTOR ELECTRONICS DECEMBER PARTS LIST Resistors: Ri, R2,R5= 5.6 KO R3=22MQ | Ra =47 ko R6, RO = 1 ko 7 = 10k RB = 1002 R10 =1.5k2 R11, R12 = 0.22 0, 25 W (see text) P1 = 100 ko preset Capacitors: C1, C3= 100 uF, 16 V, radial C2=2.2uF,25V C4 = 100 nF (0.1 uF) Semiconductors: D1, D2 = 1N4004 3 = two-colour LED D4 = 1N4148 Ti =BC557B T2=BC547B ICt =TLo81 Miscellaneous: K1, K2, K3 = 2-way terminal block for PCB mounting, 5 mm grid Ret = 12 V, 8A (car type) relay ‘S1 = single-pole push-button make ‘switch Operation When the 12 V supply ison, but no charging is taking place, D, lights orange. When 5,15 pressed to start charging, and during charg: ing, D,lights ed. Oncompletion of the charg- ; when the relay is deenergized, D; lights orange again. If something is wrong, Dydoes not light at all, of course. If the NiCd battery to be charged is con- nected with incorrect polarity, the relay will clatter for an obvious remedy, When a battery has been chai about 20 seconds (to give C; time to dis charge) before connecting the next battery to be charged to K; 1, wait Options The circuit as shown in Fig, 2 is designed for a7.2 V, 1.2 Ah NiCd battery. If batteries with a different voltage or capacity are to bocharged, R, and R, musthe suitably adapted. That requires some arithmetic, for which the stated battery properties can be used. A 7.2 V, 1.5 Ah, 6-cell NiCd battery is fastcharged at 35-4.0 A. At that current, the charging volinge is some 1.6 V per cell, Together with the drop across the connect- ing cables and contacts, a total charging voltage of some 10 V is therefore required. Since the supply is 12-13 Y, the resistors have to drop, say Ata charging cur- rent as stated, that would need a resistance value of, 2. Thisrequirementismet by two 0.33 2 resistors in series. If another type of NiCd battery, say; a 6 V, 600 mA type, w re to be charged the charging voltage would be around 8 V at a current of some 2 A. The power resistors would thus have to drop about 4.5 V; in other words, theircombined resistance would have to be 2.25 and this would be met by two 1.2 resistors in series, Their rating, P, iscalculated from P=R, where fisthe charg. ing current and R the resistance. In the case stated, the rating of each resistor would have to be 5 W .DESIGN IDEAS The contents of this article are based solely on information sup- plied by the author and do not imply practical experience by Elekor DIGITAL BLACKJACK AN EXERCISE IN SEQUENTIAL CIRCUIT DESIGN This article describes each step in the design process of a digital machine that plays the card game of blackjack (also known as vingt-et-un). The machine is a synchronous sequential circuit, the functions of which are based on the rules of By Robert F. Hodson S a teacher of digital sys- without going over 21. The lear sequential circuit design, get 10 know the dealer's play ‘ems T have found the point values for the cards are: and if the problem is too large ing strategy. This allows the game of blackjack to be an in- 10 points for face cards, to fiton paper it does not make players. to make informed structive example for individu- | point or 11 points for an ace for a good working example. choices in determining their als beginning to lear the (it is the player's choice) and Also, as it turns out, this modi- own strategy. For your game, Principles of sequential circuit the face value for all other fied’ game is sophisticated the will be as design. Choosing a game as a cards, Going over 21 is called enough to make a good compe- follows: design project makes the cir- “busting” and results in loosing tition, One more thing: as in cuit’ design fun for all the hand. The player requests regular blackjack, the players 1, Ifthe dealer has less than 16, Typically at the end of a se- another card by saying “hit mester, the students will try fo and tells the dealer to. stop beat the dealer and their fellow dealing cards by saying “stay classmates in a design contest, ‘The dealer also plays a hand, This g and the person closest to 21 is for students to implement, not the winner. In the event of a tie only a working system, but a score, the dealer wins, If you design that proves itself supe- ask a serious card player, Pam rior. In this article T will de- sure you will get a miich longer peure Combinational ‘cure: scribe the game of digital description of the game, how Girt blackjack, and then present a ever for our purposes I think typical solution to the problem. the above description is ade The solution will take the form of asynchronous sequential blackjack is a modi circuit, which will be ex- fied version of the card game. plained shortly. The design In this version, the dealer plays er's strate; jes added motivation process his a number of steps, against digital cireuit with the bur each step in. self is card strategy built in, The cir emery straightforward. After diss cuit inputs the values of the [eee cussing these seps, [will en- cards, and determines whether deavour to present some circuit another ard is ne implementations using typical hit), or to stop play r Memory digital devices. Since the actual card game Device would require a large number — What is digital of inputs to represent al the present next different type cards in a dec coe on blackjack? the modified game is played 2 Most people have played, or at with only three types of cards least heard of, the card game jacks worth 10 poinis, aces tee? blackjack. In the actual ‘card worth 1” pointsy and’ fives ‘game, the player gets one card worth 5 points. I realize that aes face down and any number of this limits the strategy of the cards face up. The object of the game considerably, but re- ime is {0 get up {0 21 points member, the objective is (0 Fig.1. Block diagram of a sequential circu. EKTOR ELECTRONICS DECEMBERhe will ake a HIT; 2.1 the dealer has 16 or more he will STAY. With an understanding of the game and the strategy in- volved, let us now clarify the concept of sequential logi Sequential circuits Before getting started on the design process, it is important that the overall concept of a se= quential circuit is understood. Digital circuits can be classi- fied as either combinational or sequential. The difference be: tween these two is that the se- quential circuit’ = can “wmember’. In mathematical terms, this means that, for a se- quential circuit, the outputs are 4 function of its present state and its inputs. Ina combina- tional circuit, the outputs are a function of the inputs only. If you are not comfortable with the mathematical description, donot worry about it. I will ex plain it another way. The pre- sent state of a digital circuit is what the circuit remembers at any given point in time, The ulputs of the circuit are de- pendent on the circuit's inputs and its present state. For exam: ple, a counter is a sequential circuit because to be able to go fo the next count value, the cir- cuit must remember its current count value. For a counter to count up to three, its present state must be two. Circuits that have outputs that depend on the current inputs and some state information are sequential circuits. The other type of digital cir cuit is a combinational circuit The outputs of combinational circuits are dependent only on, ‘the current inputs to the circuit An adder would be an example of a typical combinational cir- cuit. The output, which is the sum of the input values, only depends on the current’ input values. In generating a sum with a digital adder circuit, there is no need to remember past sums. Since there is no re- membering of past inputs in determining the sum, the adder is a combinational circuit. Figure | shows a block dia- gram of a sequential circuit Note that the circuit consists of 2 combinational circuit con nected with some memory de- ‘The memory devices ELEKTOR FLECTROD hold the present state of the system. The inputs of the « cuit are combined with present state signals (0 generate the outputs and the next state to be stored in the memory devices Let us take the counter exam- ple again, and see how it f into this general block diagram for a sequential circuit Assume the counter has one digital input called COUNT. When COUNT is a logic 1, the counter increases — otherwise, when COUNT is a logic 0, the counter holds its current value The current value of the counter is the present state of the sequential circuit, and stored in the memory devices The present state and COUNT are inputs to the combinational circuit, which is the next value in the counting sequence. ta this simple example, the out- puts of the cireuit are equal to the next state of the circuit, In eneral, this may not be the case. The outputs may depend oon both the inpuis and the pre- sent state. One characteristic of sequential circuits. worth not: ing at this point isthe feedback path from the outputs of the combinational cireuit, through the memory devices. back to the inputs of the combinational cireuit. Whenever you see a digital circuit with a feedback path like this, you ean immedi- ately classify that circuit as a Sequential circuit. The feed- back in a sequential cireuit is critical in making the circuit work. New information is stored in the memory, and is then input back into the circu This feedback mechanism is what makes a counter count from one, two, three, and so ‘One last point about sequen- tial circuits. There are actually two types of sequential circuit: synchronous and asynchro- nous. Synchronous sequential circuits are_more commonly found in systems today. The fundamental difference be- tween a synchronous. and an asynchronous. sequential cir- cuit is the presence of a clock signal. The clock is a periodic timing signal that coordinates the loading of the memory de- vices. When the memory de- viees are synchronized with the clock, the neat state ofthe ci- cuit changes at discrete inter vals in time. Synchronous sequential circuits are, in gen- Ss DECEMBER 1992 DIGITAL BLACKJACK Loo} eral, easier to design, have fewer noise problems and are less critical as regards timing. ‘The rest of this article will focus on synchronous sequen- lial circuits. The design process sign of the digital blackjack circuit. The following is a list of the steps in the design process. The list looks fairly Tong, but do not fet that deter you, because each step is Straightforward. Due to the ‘number of steps involved in the design of sequential circuits, it is not a bad idea to make your- self a check list to ensure you do not overlook a step in the deign process, Design step Completed 1. Understand the problem 2. Determine a strategy 3. Identify inputs and ‘outputs 4, Determine the state diagram 5. Make a state assignment 6. Derive the state table 7. Minimize the state table 8. Select a memory device 9, Determine next state equations 10, Determine the ourpat equations 11 Implement the circuit 0 CooDD o oo co Understanding the problem I know this is obvious. Who would design a digital circuit lo solve a problem they did not understand? Most people would not deliberately do this, but it is all too common that people proceed with # design without fully understanding what is asked for. Misunderstandings early in the design process cause costly re visions later on. In every de- sign, the bulk of the design effort should be spent at the be- ginning. This requires the de- signer to be patient. The designer should think through the problem several times be- fore proceeding. Just like pro- grammers like 10 go right to coding a solution, hardware designers love to get to a cite cuit implementation. Avoid the temptation! Think throu your design and proceed Sowly. Determining a strategy In the deseription of the digital blackjack game, the dealer's strategy is given. In this prob- lem, you need to determine @ playing strategy to ty and beat the dealer. This in itself isnot a trivial problem. Just like in the regular blackjack game, there are odds to be considered. There are a number of ways 10 determine a successful playing strategy. including statistical analysis, simulations of simply playing experience. Since the focus of this exercise is not card playing, but circuit de- sign, Twill simply state the strategy implemented in this design 1. If the point total is 16 oF Digital Blackjack Sequential Ciecuit input eseriation Xx x feardinot not vate h 0 0 |caicarie ho 1 | caraica took fi 1 0 [cers an ace 11+ |Uoused input combiner Fig. 2. input-output definitions.