ms, elektor double issue with mofe than lOO circuits summer circuits 78 pre- consonant a high performance disc preamp consonant a complete audio control amplifier luminant • novel LED level ndicator and: i bicycle speedometer • broadband rf amplifier • let miltlvoltmeter I voltage-controlled .audio mixer ■•‘disc jockey killer • brake efficiency meter • power supplies • amplifiers • test signal generators and: sone 90 other circuitsl elektor july/august 1978 - UK 5 r AUDIO AGC microphone preamp 80 amplifier for low-Z headphones ... 6 analogue delay line 1 consonant . . 53 digital audio mixer . 57 digital delay line 56 disc preamp 75 DJ killer 71 FET audio amplifier 39 ' limiter/compressor 96 luminant 54 micropower amplifier 22 1 preconsonant 52 stereo width control 33 | 18 dB per octave high/lowpass filter 27 CAR & BICYCLE automotive voltmeter 16 bicycle speedometer 50 brake efficiency meter 94 &r ammeter 87 car battery monitor 43 car voltage/current regulator 69 flasher bleeper 42 power flasher 40 DESIGN IDEAS bandwidth limited video mixer ... 92 cheap crystal filter 58 i electronic variable capacitor 62 hum filter using electronically simulated inductor 34 improved 723 supply 29 no-noise preamp 82 selective bandpass filter 48 simple CMOS squarewave generator 9 simple LS TTL squarewave generator 8 simple TTL squarewave generator 7 speedy rectifier 37 stable-start-stop-squarewave 49 super-simple touch switch 14 tern peratu re-com pensated reference voltage 99 TTL-LC-VCO 20 l iable capacitance multiplier ... 63 voltage mirror 36 Jj zero crossing detector 25 18 dB per octave high/lowpass filter 27 DOMESTIC automatic aquarium lamp 79 automatic shutter release 89 back and front doorbell 35 bat receiver 90 burglar alarm 19 cold shower detector 64 electronic gong 13 electronic soldering iron 73 infra-red receiver 21 1 infra-red transmitter 31 liquid level alarm 38 slide- tape synchroniser 102 solid-state thermostat 67 touch dimmer 100 ultrasonic alarm indicator 47 ultrasonic alarm receiver 45 ultrasonic alarm transmitter 44 GENERATORS AM/FM alignment generator 88 digital spot sinewave generator ... 4 frequency synthesiser 103 sawtooth oscillator 18 signal injector 41 simple CMOS squarewave generator 9 simple LS TTL squarewave generator 8 simple TTL squarewave generator 7 sine-cosine oscillator 11 squarewave oscillator 98 stable-start-stop-squarewave 49 TTL-LC-VCO 20 288 MHz generator 5 HF cheap crystal filter 58 cheap r.f. amplifier 10 FM IF strip 93 frequency synthesiser controller . . 95 modulatable power supply 78 VHF preamp 26 wideband RF amplifier 23 2 m transmitter 65 288 MHz generator 5 MICROPROCESSORS CMOS FSK modulator 72 data bus buffer 85 data multiplexer 86 debouncer 46 hexadecimal display 101 programmable address decoder ... 97 software Kojak siren 12 stable-start-stop-squarewave 49 supply failure indicator 3 word trigger 61 MISCELLANEOUS analogue-digital converter 32 infra-red receiver 21 infra-red transmitter 31 many hands make light work .... 104 metronome 15 pseudo random running lamp .... 76 quizmaster 70 saw-song 51 simple video mixer 74 simple video sync generator 28 squarewave-staircase converter ... 24 stereo vectorscope 68 touch controller 59 MODEL BUILDING electronic gong 13 glowplug regulator 66 model railway lighting 91 super cheapo NiCad charger 2 POWER SUPPLIES improved 723 supply 29 modulatable power supply 78 symmetrical ±15 V/50 mA supply . 17 temperature-compensated reference voltage 99 voltage mirror 36 0-30 V regulated supply 77 78L voltage regulators 81 TEST AM/FM alignment generator 88 analogue-digital converter 32 digital spot sinewave generator ... 4 DIN cable tester 84 FET millivoltmeter 60 frequency synthesiser 103 HF current gain tester 30 1C counter timebase 55 LED X-Y plotter 105 sawtooth oscillator 18 signal injector 41 sine-cosine oscillator 11 squarewave oscillator 98 timebase scaler 83 word trigger 61 REGULAR FEATURES advertisers index UK-34 linear ICs UK-18 market 7-95 missing link 7-55 selektor UK-20 TUP-TUN-DUS-DUG UK-19 STOP PRESS In the disc preamp (circuit 75) R2 is shown as 470 k. This should be 47 k. UK 20 — elektor July /august 1978 Microprocessors and memories make their mark on communications In most branches of electronics and telecommunications the key to many recent developments is the expanding use of digital techniques that employ a restricted number of simple, repetitive, coded waveforms instead of the almost infinite variations of the traditional analogue approach. The stimulus for much of this work has been the need to provide an interface with the digital computer, but the applications already extend far beyond the world of data processing. In communications the use of coded digital waveforms is not a new concept. Telegraphy was founded on the Morse code, which if it had been devised recently might be regarded as just one form of a binary, non-return-to-zero digital code. But today it is in the field of distributed and automatic control rather than in signalling that the most exciting developments are taking place, based on microprocessors and integrated circuit memories. In general, digital techniques provide the telecommunications engineer with signals that are highly resistant to noise and distortion during transmission and capable of unlimited regeneration. They thus can provide information and control signals that are easy to process, store and read out and that can be integrated with large or small computer systems. Teletext and Viewdata The ability of digital transmission systems to handle enormous amounts of data at speeds of hundreds of thousands of words a minute already has brought a new look to broadband communications systems, reaching even the consumer market in the form of the broadcast teletext services operated by the two British broadcasting authorities and the proposed interactive Viewdata system developed by the British Post Office. Such systems take advantage of the low cost of electronic memories and the associated electronic character generators by using them with domestic television receivers as visual display units, with digitalised alphanumerical and simple graphical information organised for transmission by mini- computers. A more recent development by the Independent Broadcasting Authority ( 1 ) is the Micro-ORACLE system, which has a maximum capacity of 60 pages rather than the 800 pages of the full teletext system. Micro-ORACLE is suitable for regional injection of pages into a broadcast system, or as a data display system in large operations centres. But not all communication is concerned with the transmission of high speed data streams. There still is an important role for thin line radio systems, particularly HF and VI1F radio communication systems carrying one to four voice channels, radio teleprinter traffic or manual telegraphy, both in civil and defence communications. With such systems even the smallest conventional computer seldom is justified. Computer Enhancement However, in recent months the micro- computer - a non volatile electronic memory and microprocessor — has begun to emerge as a powerful new tool for small communications systems. It provides technical operators with valuable new facilities, in much the same way as microelectronics revolutionised the world of pocket calculators. Electronic memories combined with processing — essentially ; providing the basic elements of a micro- 1 computer — are a convenient means of introducing greater flexibility rather | than complete automation and providing at relatively low cost a computer enhanced rather than a | computer controlled system. | The central processing unit of a microprocessor makes it possible for digital control circuitry to perform | routine, time consuming tasks that can lead to boredom and inefficiency among human operators. This Photo 1. The Plessey type PR 2250 communi- cations receiver is designed principally for surveillance and monitoring applications. Covering the 10 kHz to 30 MHz range, this synthesised receiver is digitally controllable from either its front panel or from an externally supplied serial data stream. An internal 16 channel memory system provides instant recall of frequencies and mode of operations. Photo 2. The Racal MA 1072 remote control unit carries all the controls and displays for the RA 1074 receiver. approach differs from earlier forms of digital control in eliminating much of the need for wired logic for each specific task. The range and scope of micro- processor applications to thin line radio communications is still evolving. But already it is clear that the incorporation into modern high performance LF/MF/HF communi- cations receivers of such a memory, based on random access and programmable read only devices, can greatly ease the routine work of the technical operator by providing a frequency-synthesised receiver with a memory of ten or more frequencies. This enables the receiver to be instantly retuned to any of the usual traffic channels by simple push button control. At any time this form of electronic memory can be up dated by means of a keypad to new channels, or used ^ to tune the receiver to any frequency. ] Such a facility may be combined with a tuning control that allows the I operator to search around the selected channel or fine tune to within a few Hertz. This technique may be extended by the use of a microprocessor to provide, for example, variable tuning rates of the I frequency synthesiser as a form of automatic electronic 'gearing’ of the I control knob, or to search automati- I cally over a given band of frequencies J until the required signal is located. Fewer Mechanical Components The control functions of modern communication receivers more and more are becoming fully electronic, with fewer mechanically controlled components. This trend, combined with the use of microprocessors to form serial data modems makes possible the complete control of receivers from central operating consoles that may be many miles from the receiver. For instance, the receivers may be located deep in the countryside far from the high levels of electrical noise found in urban areas, yet the operator can sit in front of a full receiver control panel at a convenient traffic centre in the centre of a town. Positive, fast acting control can be provided at a cost much lower than was feasible with analogue remote control techniques. Microprocessor remote control systems with master and slave units are available from several British firms. Apart from the basic remote control facility, such systems make it possible for one man at a single set of controls to operate a whole bank of distant receivers, exactly as though the receivers were mounted at his console, with such additional features as the ability to select appropriate directional antennae. High frequency radio communication, unlike microwave systems, seldom elektor july/august 1978 - UK 21 I lends itself to full automation, since I the constantly changing and largely I unpredictable ionosphere calls for I repeated changes of frequency I ^flannels that do not follow an exact I 5aily pattern. However, where a degree of automation is possible - for example in some specialised monitoring and surveillance applications - a microprocessor within the receiver can be linked into centralised computer control I systems. Continuous Coverage A number of new items of communi- cations equipment utilising these techniques have been introduced by British manufacturers. Plessey (2), for ♦xample, has demonstrated a new PR 2250 series of high performance communication receivers that give continuous coverage of the radio spectrum from 1 0 kHz to 30 MHz. The range incorporates a non volatile memory capable of instant recall of any I 16 stored frequencies, including the I appropriate mode, bandwidth and I automatic gain control characteristics. I A keypad not only allows the operator I to enter and up date the memorized information, but lets him change immediately to any frequency without touching the tuning knob. Even so, the tuning control then is available immediately for the operator to search around the new frequency with a tuning rate of either 20 or 1 kHz with 10 Hz increments - per knob revolution. Although the front panel of the PR 2250 is not unlike a traditional communications receiver - apart from the provision of a keypad - there are no mechanical linkages to the main receiver circuits, which are in the form of demountable modules. The electrical connections are made via flexible printed wire strips, so that the receiver quickly can be broken down into a | series of building blocks. This reduces the mean time to repair since a fault can I be cleared by simply plugging in a replacement module. In one screened module is a micro- | processor unit that allows the receiver to be readily connected to the various forms of programmed control that are becoming a feature of modem monitoring systems. Synthesised Remote Control Clearly such digital techniques are best applied to a receiver capable of meeting the highest requirements in respect of dynamic range, frequency stability and low susceptibility to spurious responses or reciprocal mixing. Associated models are available providing full remote control. Another company that has developed receiver systems featuring advanced digital techniques is Racal Communi- cations (3). Its current range includes the RA 1 784 synthesised remote control receiver for use in conjunction with the MA 1072 Score unit. Score is an acronym for Serial Control of Racal Equipment. Score provides a distant operator with a control panel virtually identical to that of the standard receiver and capable of controlling all the variable facilities of the receiver. These include not only frequency, mode and bandwith selection but such analogue functions as intermediate frequency gain and beat frequency oscillator tuning. The system operates by scanning each of the parallel control wires in the MA 1072 unit in sequence and assembling the data into a composite serial 48 bit data frame with each controlled parameter allocated a number of data bits. In more complex installations up to 14 receivers may be controlled by a minicomputer supported by a floppy disc store and a visual display unit. Flexible System The availability of low cost visual display units, keyboard terminals and microprocessors raises many worthwhile possibilities in control, information retrieval and display systems for incorporation in communication systems. A flexible digital remote control system, the H 6800, has been developed by Marconi Communication Systems (4) to allow remote control of complex communications facilities over a single telephone line. Basically it comprises a visual display unit with keyboard, with an associated micro- processor unit, and it readily can be applied to a diverse range of control and monitoring requirements. Further exploitation of the processor can provide a degree of automation of the complete system. For instance, modern fast tuning HF transmitters could be used as radio sounding systems with control and evaluation by means of the processor to provide feedback of the ionospheric conditions. Similarly it could take over the organization of secure frequency agile operation, day/ night frequency selection, antenna selection for specific circuits, transmitter fallback systems, and even routine logging operations. The planning of such largely automated and unattended systems is a striking illustration of the impact of digital techniques and the application of micro and minicomputers to HF communi- cations. And the marriage of low cost computing techniques with communi- cations rapidly is opening up more possibilities, ranging from simple operator aids to highly complex automatic systems. Key ( 1 ) IBA Engineering Centre, Crawley Court, Winchester, Hampshire, England. (2) Plessey Avionics and Communi- cations, Martin Road, West Leigh, Havant, Hampshire, England. (3) Racal Communications Ltd, Western Road, Bracknell, Berkshire, England. (4) Marconi Communication Systems Ltd, Marconi House, Chelmsford, Essex, England. UK 22 - elektor july/ai 1978 bias components are incorporated into the input of the buffer, so the input signal must be AC coupled (capacitor or transformer) unless response down to DC is required. Pin 17 is provided to allow biasing of the crossover point of the input buffer. Crossover distortion can be minimised by sinking up to 100 pA out of this pin to a more negative point. The input balance point, pin 19, allows control voltage feedthrough to be minimised by either sinking or sourcing up to 100 pA at this pin. The simulated potentiometer also has taps on the ‘top’ and ‘wiper’ which are impedance controlled and allow adjustment for maximum linearity. The output stage will drive loads down to 47 k in parallel with not more than 100 pF, although lower impedances The test pin 4 is approximately one base-emitter voltage drop above the negative rail and should be left floating during normal use. The electrical specifications of the MC 675 are given in table 1 . whilst the mechanical dimensions are given in figure 2. Cadac (London) Ltd., 141, Lower Luton R&zd. Harpenden, Hens A1 (338 S) I analogue delay line 3 There are numerous applications requiring the use of an audio delay line, for example phasing and vibrato units, echo and reverberation units, and sophisticated loudspeakers with active time-delay compensation. One of the simplest ways to achieve this electronically is to use an analogue (bucket brigade) shift register. There are various types on the market and a particularly interesting one is the Reticon SAD 512, which has 512 stages and a built-in clock buffer. The clock buffer enables it to be „ driven from a simple, single-phase ‘ clock circuit such as a CMOS multi- vibrator. Figure 1 shows a delay line utilising the SAD 5 1 2. Input signals to the device must be positive with respect to the 0 V pin, so the AF input signal is first fed to an inverting amplifier, IC2, which has a positive DC offset adjustable by PI . The clock generator is an astable multivibrator using CMOS NAND gates (N1 and N2) and its frequency may be varied between 10 kHz and 1 00 kHz by r means of P3. The clock buffer of the SAD 512 divides the clock input frequency by two, so the sampling frequency, f c , of the SAD 512, varies between 5 kHz and 50 kHz. The delay produced by the circuit is n/2f c , where n is the number of stages in the IC. The delay may therefore be varied between 5.12 ms and 5 1 .2 ms. To obtain longer delays several SAD 5 1 2s may, of course, be cascaded very easily, since no special clock drive circuit is required. . v To minimise clock noise the outputs from the final and penultimate stages of the IC are summed by R7, R8 and P2. However, if the circuit is to be used with the minimum clock frequency then clock noise will still be audible, and the lowpass filter circuit shown in figure 2 should be connected to the output. This consists of a fourth-order Butterworth filter with a turnover frequency of 2.5 kHz and an ultimate slope of -24 dB/octave. Of course, if low clock frequencies are to be employed then the maximum frequency of the input signal must be restricted to half the sampling frequency. This can be achieved by connecting the filter circuit of figure 2 at the input of the delay line, as shown in figure 3. To set up the circuit the clock frequency is lowered until it becomes audible. P2 is then adjusted until clock noise is at a minimum. The clock frequency is then raised and a signal fed in. The signal level is increased until distortion becomes apparent, whereupon PI is adjusted to minimise distortion. This procedure (increasing the signal level and adjusting PI) is repeated until no further improvement is obtained. Alternatively, if an oscilloscope is available, PI may be adjusted so that the waveform clips symmetrically when the circuit is overloaded by a large signal. elektor july/august 1978 — 7-01 super cheapo NiCad charger Nickel-Cadmium rechargeable batteries are gaining popularity in many applications, as they offer (in the long term) significant savings over dry (primary) batteries. Of course, the initial outlay involved is increased because a charger unit is required; however, this simple charger can be built using com- ponents that may be found in almost any constructor’s junk box. For maximum life (number of charging cycles) NiCad batteries should be charged at a fairly constant current. This can be achieved quite simply by charging through a resistor from a supply voltage several times greater than the battery voltage. Variation in the battery voltage as it charges will then have little effect upon the charge current. The circuit consists simply of a transformer, diode rectifier and series resistor as shown in figure 1 . The accompanying nomogram allows the required series resistor value to be calculated. A horizontal line is drawn from the transformer voltage on the vertical axis until it intersects the required battery voltage line. A line dropped vertically from this point to intersect the horizontal axis then gives the required resistor value in ohms. As an example, the dotted line in figure 2 shows that if the transformer voltage is 18 V and a 6 V battery is to be charged then the required resistance value is 36 ohms. This resistance value is for a charging current of 1 20 mA and if other values of charging current are required the resistor value must be scaled accordingly, e.g. 1 8 ohms for 240 mA, 72 ohms for 60 mA etc. D1 2 J may also be replaced by a bridge rectifier, in which case the resistance value for a given current must be doubled. The power rating (watts) of the resistor should be greater than 1 2 R, where I is the charge current in amps and R the resistance in ohms. As the circuit does not incorporate any form of charge cutoff the charge rate must not be too great or the life of the battery may be reduced. As a general rule it is permissible to charge most NiCads at a current of 0.1C or less for several days, where C is the capacity of the battery in ampere-hours. supply failure indicator Many circuits, especially digital systems such as random access memories and digital clocks, must have a continuous power supply to ensure correct operation. If the supply to a RAM is interrupted then the stored information is lost, as is the time in the case of a digital clock. The supply failure indicator described here will sense the inter- ruption of the power supply and will light a LED when the supply is restored, thus informing the microprocessor user that the information stored in RAM is garbage and must be re-entered, and telling the digital clock owner that 7-02 — elektor july/august 1978 his clock must be reset to the correct time. When the supply is initially switched on the inverting input of IC1 is held at 0.6 V below positive supply by D1 . Pressing the reset button takes the non-inverting input of IC1 to positive supply potential, so the output of IC1 swings high, holding the non-inverting input high even when the reset button is released. LED D2 is therefore not lit. When the supply is interrupted all voltages, of course, fall to zero. Upon restoration of the supply the inverting input of IC1 is immediately pulled up to its previous potential via D1 . However, Cl is uncharged and holds the non-inverting spat low, so the output of IC1 rerruams low and D2 lights. digital spot sinewave generator Low-distortion spot-frequency sinewave generators are extremely useful for carrying out distortion measurements on audio equipment. Unfortunately most analogue cir- cuits, which rely on thermistors or FETs for amplitude stabilisation, suffer from amplitude bounce due to the long time constant of the stabiliser circuit, which is necessary to achieve low distortion. By synthesising a sinusoidal waveform digitally the problem of amplitude instability can be avoided. The circuit consists principally of a clock oscillator built around N1 and N2, a divide-by-32 counter IC3, and a 16-bit serial-in-parallel-out shift register comprising IC 1 and IC2. The Q5 output of IC3 is connected to the Data input of IC1. For the first 1 6 clock pulses this output is high, so ones are loaded into the shift register and clocked through until all 16 outputs are high. Each output of the shift register is connected to P2 via a resistor, so as the shift register outputs go high the voltage across P2 varies in a series of steps. By suitable choice of resistor values the waveform thereby appearing across P2 can be made to be half a Resistors: R1 = 39 k R2 = 8k2 R3,R30,R31 ,R37 = 1 M R4.R10.R22.R28 = 6k8 R5.R29 = 330 k R6,R26 = 27 k R7.R27 = 180 k R8.R24 = 5k6 R9.R25 ■ 150 k R11.R23- 120 k R12.R20 = 12k R13.R15.R16.R17, R19.R21 = 100 k R14.R18 = 2k2 R32.R33 = 18 k R34.R35 = 33 k R36= 10 k PI = 10 k preset potentiometer P2 = 4k7 (5 k) preset potentio- C1 ,C2,C3 = see text C4 = 10 p/25 V C5 = 680 n C6= 100 n C7 = 270 n Semiconductors: IC1 ,IC2= 4015 IC3 = 4024 IC4 = 4069 IC5.IC6 = LF 356 cycle of a sinewave, from trough to peak. Since all four quadrants of a sinusoidal waveform are symmetrical it is a simple matter to synthesise the other half-cycle of the waveform, from peak to trough. From the 17th to the 32nd clock pulse the Q5 output of 1C3 is low, so the shift register is loaded with zeroes and the voltage across P2 falls back to zero in an exact mirror image of its rise to the peak. On the 33rd clock pulse the Q5 output of IC3 again goes high and the whole cycle repeats. The waveform across P2 contains a large proportion of clock frequency components and also harmonic distortion due to the resistor values not being exactly correct, so these unwanted components are removed by a filter constructed around IC6. P 1 is used to trim the clock oscillator frequency so that the frequency of the output sinewave is exactly at the centre of the filter’s response. To do this the output from the filter is measured on an AC voltmeter and P3 is adjusted until maximum output is obtained. P2 may then be used to adjust the output level between zero and 6.5 V peak-to- peak. elektor july /august 1978 — 7-03 Using only a small number of TTL gates it is not difficult to construct a squarewave generator which can be used in a wide range of possible applications (e.g. as clock generator). The accompanying diagram represents the basic universal design for such a generator. The circuit is not critical, can be used over a wide range of frequencies, has no starting problems and is sufficiently stable for most applications. The frequency in not affected by supply voltage variations. The oscillator frequency is determined by the RC network and the propagation time of the inverters (in this case three NANDs with their inputs connected in parallel). Since the propagation delay time of the IC is, in general, strongly influenced by the temperature and supply voltage, care must be taken to ensure that the propagation times have as little effect as possible upon the oscillator frequency. The output of each gate changes state twice per period of the oscillation signal, which means that, in all, one must account for double the propagation time of all three gates. To ensure that the oscillator frequency f 0 be more or less The design of this squarewave generator is identical to that of the circuit described above, with the exception that it employs an IC from the increasingly popular Low Power Schottky (LS) TTL series, rather than a conventional TTL 1C. Since the electrical characteristics of LS TTL devices differ from those of standard TTL ICs, the relationship between the oscillator frequency and the values of R and C will also be difficult, whilst an extra resistor is required for the circuit to function satisfactorily. The circuit will generate a square- wave with a frequency between 20 Hz and 1 MHz. The nomogram once again shows the frequencies obtained for various values of R and C. As was the case in the above circuit, there is a minimum permissible value for R (680 S2). To obtain a variable squarewave generator, R should be replaced by a 5 or 1 0 k potentiometer in series with a fixed resistor of 680 £1. 7-06 — elektor juiy/august 1978 simple TTL squarewave generator ic "5 s>-2— T°' independent of variations in the temperature of the circuit and in the supply voltage, one must ensure that f 0 is small compared to ? , 2 • t p • n where tp is the propagation time and n the number of inverters connected in series. In the case of the circuit shown here, tp = approx. 10 ns and n = 3, so that as far as the oscillator frequency f 0 is concerned: f » < 2T7~-27Tkl-‘ 6 - 6MH2 - The accompanying nomogram shows how f Q changes with R. The value of the resistor R must not be smaller than that shown in the nomogram; for example, for C = 100 nF, R must not be less than 1 00 S2. A variable squarewave oscillator can be obtained by replacing R with a 2 .5 k potentiometer in series with a fixed resistor of the minimum permitted A universal squarewave generator of this type can also be constructed using Low Power Schottky TTL or CMOS ICs; both these possibilities are discussed below. simple LS TTL squarewave generator IC “ m jjjlpfcJiim : : K j Hi :eee — — vflUn -5 -MMM — - -- Jj = =z -MM EE: \ ill L simple CMOS squarewave generator In addition to standard- and Low Power Schottky TTL devices, there is, of course, no objection to using a CMOS IC in the basic squarewave generator circuit , The revised graph of frequency against R and C is shown in the accompanying nomogram. The frequencies are plotted for a nominal supply voltage of 12 V, however, this voltage is not critical, and a supply of between 5 V and 15 V may in fact be used. The frequency range of the circuit runs from 0.5 Hz to 1 MHz. » The minimum permissible value of R is 22 k. To obtain a variable squarewave oscillator, R should be replaced by a fixed value resistor of 22 k in series with a 1 M poten- tiometer. Both the buffered and unbuffered versions of the 401 1 may be used. ic £ 1 T cheap r.f. amplifier 0 It is possible to reduce distortion in r.f. amplifiers by employing negative feedback. When using this technique, however, it is important that the feedback network does not cause mismatching between input and output, which basically means that transformer coupling should be used. Using this approach it is possible to accurately match the input and output impedances whilst also achieving a low noise figure. A prototype model which used a BF 199 gave the following results: bandwidth (—3 dB): 0.11 -40 MHz gain: 1 1 dB two-tone test: P ou t = +10 dBm/tone, third-order IM - 40 dB with respect to one output signal, noise: 15 dB The relatively poor noise figure is due to the type of transistor which was used. Better results can be obtained using CATV transistors (BFW 30, BFW 16, BFR 94, BFR 65, BFR 64 etc.) which give a low noise figure despite their (relatively) high collector currents. Lit: CQ DL 2/78 pp. 64,65 elektor july/august 1978 - 7-07 sine-cosine oscillator There are a number of applications which require two sinewave signals that are of the same frequency but 90° out of phase, i.e. a sine signal and a cosine signal. Such signals are used in SSB and quadrature modulation, electronic generation of circles and ellipses and transformations between rectilinear and polar coordinates. Sine and cosine signals can be obtained from a quadrature oscillator which consists of two integrators connected as shown. A1 is connected as a non-inverting integrator, while A2 is connected as an inverting integrator. Why this circuit should produce a sine and cosine signal may not be immediately apparent, but is easily explained. At output B appears a signal which is a function of time, f(t). Since this is minus the integral of the signal at A it is obvious that the signal at A is minus the differential of the signal at B i.e. — Similarly, the input dt signal to integrator A 1 is the differential of the signal at A, i.e. However, the input signal to A 1 is the output signal from A2, i.e. conditions are satisfied by sine and cosine signals, since, if f(t) = sin co t (output B) I , i l ( rl — — = cos co t (output A) dt d (cos co t) _ d 2 (sin co t) _ dt dt 5 Sm ° = -f(t) Output A therefore produces a cosine signal and output B a sine signal. PI is used to adjust the loop gain of the circuit so that it oscillates reliably. If, due to component tolerances, the circuit does not oscillate at any setting of P 1 , it may be necessary to increase its value to 10 k. The signal amplitude is stabilised by D1 . D2 and R4 to R7. The frequency of oscillation can be altered by substituting different values of capacitor for Cl to C3, calculated using the equation given. Literature. Texas Instruments Application Notes. software Kojak siren The number of possible applications for the SC/MP microprocessor are legion, however thanks to the DELAY instruction it is particularly easy to use ‘SCAMP’ to generate low frequency signals. The following programme for a police siren is a good example of this facility. A periodic signal is obtained by repeatedly setting and resetting a flag. The siren effect is produced by using DELAY instructions to vary the interval between set and reset. As it stands, the accompanying programme will generate a double- siren effect, similar to that of American police cars. If a ‘normal’ siren is desired, the contents of 7-08 — elektor july/august 1978 address 0F12 should be altered to 90. The signal is rendered audible by means of the loudspeaker interface shown. This is an extremely simple little circuit which is connected to flag 1 of the SC/MP. The volume is controlled by means of PI . This circuit will simulate the sound of a bell or gong and may be used as a replacement for conventional bells in such applications as doorchimes, clocks etc. The circuit consists of a resonant filter built around IC2 and IC3 which will ring at its resonant frequency when a short pulse is fed to the input. In this circuit the trigger pulses are provided by a 555 timer connected as an astable multivibrator, ► but other trigger sources may be used depending on the application. The character of the sound is influenced by two factors; the Q of the filter, which may be varied by changing the value of R2, and the duration of the trigger pulse, which may be adjusted using PI . The repetition rate of the trigger pulses, i.e. the rate at which the gong is ‘struck’, may be varied using P2. In order to drive a loudspeaker the output of the circuit must be fed through an audio amplifier. The output level may be varied from zero to about 5 V by means of P3. super-simple touch switch Although there is a plethora of designs for touch switches, it is always a challenge to come up with a design that is simpler than previous versions. While most latching touch switches use a pair of NAND gates connected as a flip-flop, this circuit uses only one non-inverting CMOS buffer, one capacitor and a resistor. When the input of N1 is taken low by bridging the lower pair of touch contacts with a finger, the output of N1 goes low. When the contacts are released the input of N1 is held low by the output via R1 , so the output remains low indefinitely. When the upper pair of contacts is bridged the input of N1 is taken high, so the output goes high. When the contacts are released the input is still held high via R 1 , so the output remains high. elektor july/august 1978-7-09 metronome B. v.d. Klugt Those of our readers who have experienced the pleasures of piano lessons during their childhood, will doubtless be all too familiar with the sound of a metronome. This is a clockwork instrument with an inverted pendulum, which can be set to beat a specific number of times per minute, the loud ticking thereby indicating the correct speed at which the passage of music should be played. Although mechanical metronomes are still used almost universally, it is, of course, also possible to achieve the desired effect electronically. The circuit for an electronic metronome described here is distinguished, not by any revol- utionary new features, but by its extreme simplicity and excellent stability. N1 to N3 form an astable multivibrator. By means of PI , the frequency of its output signal can be varied between 0.6 and 4 Hz, whilst the pulse width can be adjusted be means of P2. The latter control modifies the sound of the beat between a short ‘dry’ tick and a longer, fuller tone. The volume control is provided by P3, which varies the peak current through the loudspeaker between 0.5 A and 50 mA. At low resistance values of P3, the resultant large current places fairly heavy demands upon the transistor, and hence a Darlington pair was chosen. Thanks to the low duty cycle, the average current drawn by the circuit is extremely small, so that an ordinary 4.5 V battery will suffice for the power supply. The accompanying photo shows how the characteristic shape of the metronome can still be conserved in the electronic model. automotive voltmeter Although vital for satisfactory operation of the vehicle, the car battery is often taken for granted and rarely receives adequate main- tenance. As a battery ages, its ability to store charge for long periods gradually decreases. The inevitable result is that one morning (usually in the depths of winter) the car fails to start. The solid-state voltmeter described in this article allows continuous moni- toring of the battery voltage so that incipient failure can be spotted at an early stage. The circuit will also indicate any fault in the car voltage 7-10 — elektor july/august 1978 ' regulator which may lead to over- charging and damage to the battery. Battery voltage can, of course, be measured using a conventional moving-coil voltmeter. However, as only the voltage range from about 9 to 15V is of interest, only the top third of the scale of a 1 5 V meter would be used, unless a ‘suppressed zero’ facility was added. Moving coil meters are also fairly delicate mech- anically. A better solution is to use a solid- state voltmeter which will indicate the voltage on a column of LEDs. Various ICs are available which perform this function. However, the 1 2 or 16 LED display offered by ICs such as the Siemens UAA 180 and UAA 170 is not required in this application, so an IC was chosen which will drive only five LEDs, the Texas SN 1 6889P. This IC provides a thermometer-type indication. The complete circuit of the volt- meter uses only this IC and a handful of other components, since the IC will drive the LEDs directly. Diodes D1 and D2 provide protection against reverse polarity and tran- sients on the supply line, whilst D8 offsets the zero of the meter so that it only begins to read above about 9.5 V. The circuit is calibrated using PI so that the LEDs extinguish at the voltages shown in the accompanying table. LED D7 will be extinguished below about 1 5 V. If this LED is lit when the circuit is fitted in the car then the charging voltage is too high and the car voltage regulator is at fault. A red LED should be used for D7. D6 indicates that the battery is fully charged, and a green LED should be used for this component. D5 indicates that the battery voltage is fairly o.k., but the battery is not fully charged - a cautionary yellow LED can be used here. D4 and D3 indicate that the battery voltage is unacceptably low, and red LEDs should again be used for these components. Table. Voltages below which the LEDs extinguish D7 15 V D6 1 3.5 V symmetrical 15 V/50mA supply Although IC voltage regulators have now largely displaced discrete component designs, this circuit offers a considerable cost advantage over an IC regulator. In fact the component cost of the regulator section is only about 75 p! Operation of the circuit is extremely simple. The centre-tapped transformer, bridge rectifier and reservoir capacitors C 1 and C2 provide an unregulated supply of about ± 20 V. The positive and negative regulators function in an identical manner except for polarity, so only the positive regulator will be described in detail. The positive supply current flows through a series regulator transistor T1 . 1 5 V is dropped across zener diode D5, the upper end being at about +1 5 V and the lower end at 0 V. Should the output voltage of the regulator tend to fall then the lower end of D5 will fall below 0 V and transistor T3 will draw more current. This will supply T1 with more base current, turning it on harder so that the output voltage of the regulator will rise. If the output voltage of the regulator is too high then the reverse will happen. The potential at the lower end of D5 will rise and T3 will draw less current. T1 in turn will tend to turn off and the supply voltage will fall. The negative supply functions in a similar manner. Since D5 and D6 receive their bias from the output of the supply, R5, R6 and D7 must be included to make the circuit self-starting. An initial bias of about 10 V from the unregulated supply is provided by these components. Once the output voltage of the supply has risen to its normal value D7 is reverse-biased, which prevents ripple from the unregulated supply appearing on the output. Using inexpensive small-signal transistors such as the BC 107/ BC 177 family or equivalents, the maximum current that can safely be drawn from the supply is about 50 mA per rail. However, T1 and T2 may be replaced by higher power Darlington pairs to obtain output currents of 500 mA. elektor july/august 1978 — 7-1 1 obtain the long period required. The output from the monostable is used to turn on T6, causing the relay to » pull in. The sensitivity of the circuit can be set by means of P2. A range of up to 10 m (30 feet) can be achieved, which should prove adequate for most purposes. It should, however, be noted that this type of alarm system is notorious for ‘false alarms’ if the sensitivity is too high. Not only flies or moths may be detected, but even a sudden draught caused by an open window! Apart from the sensitivity, which of course can be set according to personal taste (if not bitter experience), the only other calibration point is PI . This sets the operating point of the Gunn oscillator. Initially, PI should be set to maximum resistance. The top end of the Gunn oscillator is temporarily disconnected from point A in the circuit, and a multimeter is connected in series. The test leads should not be unnecessarily long, and they should be tightly twisted. The meter is set to a suitable current measuring range and PI is adjusted until the current consumption of the oscillator is 120 mA. After re- connecting the oscillator to point A the circuit is ready for use. As a final note, R17, T6, D3 and the relay will not always prove necessary: either R 1 1 or R16 may be replaced by a reed relay if only a low alarm current is to be switched. Philips/ Mullard Application notes for CL8630S Note: In the UK Home Office type approval must be obtained for microwave intruder alarms. TTL-LC-VCO LI = 18 turns 21 SWG tl mm) enamelled copper wire on a 3 mm dia. former This Voltage Controlled Oscillator (VCO) uses only two TTL gates or inverters, so it may prove useful in ^ digital circuits where one or more J TTL ICs are not fully utilised. Any type of TTL gate that can be wired as an inverter can be used (NANDs, NORs or inverters). Basically, the circuit is an extended version of the well-known two-gate RC oscillator. However, in this case the main frequency-determining element is an LC resonant circuit consisting of LI , C2 and D1 . Since D1 is a varicap, a control voltage applied to R3 will alter the resonant frequency of the circuit. A control voltage range of 0 ... 5 V corre- sponds to an oscillator frequency range of 7.5 .. . 9.5 MHz. The output is, of course, TTL compatible. With the component values shown, the circuit is only suitable for TTL gates: for one thing, the frequency is too high for CMOS. However, the same principle can be used with CMOS provided the circuit is re- designed . The performance of the circuit is not particularly good - the linearity, for instance, is mediocre — but it is reliable and cheap. One possible application is where a clock frequency is required that can be varied as a function of logic states elsewhere in the circuit. elektpr july/august 1978 — 7-13 infra-red receiver ^ "4 I h® .N T1.T2.T3 = BF 494, BF 194 T4.T6.T7 = BC 557B, BC 177B T5 = BF 256A.B This IR receiver can be used with the complementary transmitter described elsewhere in this issue. The IR signal is received by photodiode D1 . This diode is reverse-biased (its bias voltage being decoupled from noise on the supply line by R4. R5, C5 and C6) and its leakage current varies with changes in incident light. To enable the receiver to be sensitive without being interference-prone it must be made selective, so a tuned circuit Ll/Cl is included. The bandwidth of the receiver is 100 Hz when tuned to 24 kHz. It is not possible to reduce the bandwidth further since the tuning is affected by the capacitance of D1 , which is light dependent. If the bandwidth were too narrow the receiver could drift off tune due to this effect. T1 and T2 form a cascode amplifier with negative feedback, whilst T3 and T4 provide further gain. The amplified signal is then detected by D2 and D3, and the resulting DC voltage is further amplified by T5, T6 (which drives an indicator LED, D4) and T7, which energises a relay. By a slight modification to the stage 7-14 — elektor july /august 1978 around T6 it is possible to link the receiver to the ultrasonic alarm indicator described elsewhere in this issue to make an infra-red alarm system. T7 and associated components may then be omitted (see figure 2). To align the transmitter and receiver the following procedure should be adopted: 1 . Switch on the transmitter and check that it is drawing a current of between 50 and 100 mA. 2. Set C3 of the transmitter circuit to its mid-position (vanes half- closed) then switch off the transmitter. 3. Turn the wiper of PI fully towards R8 and the wiper of P2 fully towards R15. The LED, D4, should now light, which indicates that the first stage of the receiver has begun to oscillate. 4. Adjust P2 until the LED only just lights. 5. Adjust PI until the LED extinguishes. 6. Switch on the transmitter and move the transmitter towards the receiver until D4 begins to flicker. Adjust Cl of the receiver until D4 glows continuously. Increase the distance between transmitter and receiver until D4 again begins to flicker and readjust C 1 . Repeat until maximum range is obtained. It may be that the transmitter frequency is outside the receiver tuning range, in which case the value of C3 may need to be changed slightly to bring the transmitter frequency within the adjustment range of Cl . Using the LD241/1 in the transmitter and BPW34 in the receiver it should be possible to obtain an operating range of at least 10 metres without any special optical system or shielding of the photodiode from ambient light. If a simple lens system and shielding tube for the BPW34 is employed then much greater ranges may be achieved. The receiver should operate from a stabilised 1 2 V supply capable of of supplying 12 mA plus the rated relay current. Any of the common 1 2 V IC regulators should prove suitable for this task. (see circuit 31) »» Sr ® ®a> 2 ® micropower amplifier This amplifier has been specially designed for use with ‘alternative energy sources’ such as solar cells, biological fuel cells, etc. These types of energy source characteristically have a low and variable output voltage and a high output resistance. The amplifier will operate reliably from supply voltages between 3 V and 20 V having source resistances as — — (Sa)'"« The power which the amplifier can supply is, of course, dependent on the supply voltage and its source resistance, as can be seen from the accompanying table. The quiescent current consumption of the amplifier is between 1 mA and 1.5 mA, the exact value depending on the type of transistors used. Should the quiescent current fall outside this range it will be necessary to vary the value of R9. As is apparent from the table, the amplifier performs best with high impedance loudspeakers. As speakers with impedances as great as 200 ohms are not readily obtainable, the alternative is to use a lower impedance speaker with a matching transformer. For example an 8 ohm speaker could be used with a trans- former having a ratio of approxi- mately 5:1. Although the output level of the amplifier is not exactly earsplitting, it is nonetheless sufficient when used with a reason- ably efficient loudspeaker in a quiet room. The voltage gain of the amplifier is approximately 50 and the 3 dB bandwidth is about 300 Hz to 6 kHz. elektor july/august 1978 — 7-15 wideband RF amplifier This design for an RF amplifier has a large bandwidth and dynamic range, which makes it eminently suitable for use in the front end of a shortwave receiver. The design operates without negative feedback since, if an amplifier with feedback overloads, distortion products can be fed back to the (aerial) input via the feedback loop and re-radiated. However, good linearity is achieved by employing a device which has an inherently linear transfer characteristic, in this case a dual-gate MOSFET with both gates linked. With the 3 N21 1 used in this circuit the transconductance of the device is constant at about 14 mA/V provided the drain current is greater than approximately 12.5 mA. The MOSFET is used in a Common- Table 1 Typical characteristics of the RF amplifier gain: approximately 10 dB 3 dB bandwidth: 4 MHz to 55 MHz noise figure: less than 5 dB two-tone test: output power for third order IM distortion at —40 dB with respect to one tone: +22 dBm/tone 20 V gate configuration, with PI used to set the drain current at around 20 mA. One home-made inductor is employed in the circuit, L2, which is wound on a Philips/Mullard type 4312-020-3 1521, two-hole ferrite bead, sometimes referred to as a ‘pig’s nose ferrite bead’. 14 turns of 31 SWG (0.3 mm) enamelled copper wire are wound through one hole of the bead and four turns are wound through the other hole, one end of each winding being joined to form the tap which connects to C6. PI should be adjusted so that the voltage at the test point shown in figure 1 is between 17.5 V and 18 V. squarewave- staircase converter R1 . . . R23 =10 k/1% 4 This circuit can be used to generate an up-down staircase waveform with a total of 5 1 2 steps per cycle. IC1 and IC2 are two four-bit up-down counters connected as an 8-bit counter, with an R-2R D-A ladder network connected to the outputs to 7-16 — elektor july/august 1978 convert the binary output codes to a staircase waveform. When a squarewave is fed to the clock input the circuit will count up until the counter reaches 255, when the carry output will go low and clock FF1 . The circuit will then count down until zero is reached, when the carry output will again clock FF1 and so on. To ensure that the staircase steps are of equal height 1% tolerance resistors should be used for R1 to R23. zero crossing detector This circuit will detect precisely the negative-going zero-crossing point of an AC waveform, but requires only a single supply voltage, unlike zero crossing detectors using op-amps. N1 and N2 are Schmitt triggers connected to form a monostable multivibrator with a period of about 15 ms. PI is adjusted so that when the input voltage falls to zero the voltage at the input of N1 is equal to the low-going threshold of the Schmitt trigger. The output of N1 thus goes high and the output of N2 k goes low. Cl holds the second input of N1 below its positive-going threshold for about 1 5 ms, during which time the output of the circuit will remain low, even if noise pulses on the input waveform should take the first input of Nl high. When the input signal goes positive the first input of N 1 is taken above its positive-going threshold. Note that this occurs after the positive- going zero-crossing point due to the hysteresis of the Schmitt trigger. Subsequently the second input of Nl goes high due to Cl charging through R3. The circuit then resets and the output of N2 goes high. The output of N2 is thus an asymmetrical squarewave whose negative-going edge occurs on the negative-going zero-crossing point of the input waveform and whose positive-going edge occurs sometime during the positive half-cycle of the input waveform. The negative-going edge of the waveform is independent of the amplitude of the input signal and occurs always at the zero- crossing point. However, it does vary slightly with supply voltage, so this should be stabilised. If a higher supply than 15 V is used then R4 and D2 must be included, otherwise the IC may be damaged. To calibrate the circuit an oscilloscope is desirable so that PI may be set exactly for the zero- crossing point. Alternatively, if a ’scope is not available, Cl should be temporarily disconnected and the output of N2 monitored on a multimeter. In table 1 look up the voltage corresponding to the RMS input voltage and supply voltage and adjust PI until this voltage registers on the meter, e.g. with a 10 V supply and a sinewave input of 5 V RMS PI should be adjusted until the meter reads 4.47 V. R1 , D1 and the input protection diodes of Nl protect the circuit against input voltages up to 220 V (RMS, sinewave input). At this level the maximum permissible current of 1 0 m A flows into N 1 and 1 .5 W are dissipated in R 1 . If higher input voltages are to be used or less dissipation is desirable then the values of Rl, R2 and PI should be increased, keeping them in the same ratio. (RMSs^ne) 5 V VHF preamp Designing a preamp for the VHF waveband (around 100 MHz) is not always an easy matter. This circuit, however, is both relatively simple to use and is inexpensive. It has the advantage of a fairly large bandwidth (2 MHz) and good noise figure (2.5 dB). The preamp has a large dynamic range and a gain of 20 dB at a frequency of 144 MHz. L 1 and L2 are air-cored coils with an internal diameter of 6 mm and consist of 4 turns of 1 mm silver- plated copper wire. LI is tapped one turn from the earthy end, whilst L2 has a tap one turn from the end nearest R3. Ceramic types are recommended for the four 1 n capacitors. elektor july/august 1978 - 7-11 18 dB per octave ^// high/lowpass filter Calculating the correct RC values for high- and lowpass filters can be something of a chore and is often regarded by amateur constructors as a subject that is best left well alone. This is especially true the more complicated the filter and the steeper the slope of the filter becomes. That being the case, the following circuit for a third order (i.e. with a slope of 1 8 dB per octave) high/lowpass Butterworth filter, together with the accompanying nomogram which supplies the correct RC values for any given turnover point, should prove extremely useful. The circuit shown is for a lowpass filter, however by changing over the position of the resistors and capacitors a highpass filter is obtained. The beauty of the circuit lies in the fact that all the resistors and capacitors have the same respective value. Either operational amplifiers or emitter followers may be used as voltage followers. Normally, to find the turnover frequency of a filter (i.e. the point at which the output voltage of the filter is 3 dB down on the passband response) one uses the equation fo (the turnover point) = 1 1(2 n RC). However one can forget about having to go through these calculations by using the accompanying nomogram. The turnover points are displayed along the horizontal axis, whilst the corresponding values for C are shown along the vertical axis. Furthermore, a number of resistance values are also indicated by the diagonal lines running across the nomogram. To use the nomogram one first draws an imaginary vertical line through the desired turnover point. An imaginary horizontal line is then drawn through the point at which the vertical line intersects with the desired resistance. The intersection of that horizontal line with the x-axis gives the correct value of C for the chosen turnover frequency. In the example shown (in dotted lines), a turnover frequency of 720 Hz is obtained with R = 1 0 k and C = 22 n. simple video sync generator This simple circuit will generate 1 5625 Hz and 50 Hz line and field sync pulses for video applications. A clock signal from a 125 kHz astable multivibrator is divided down by a 4040 12-bit counter, the Q outputs of the counter being NANDed together to give line pulses with a duration of 4 /as and field sync pulses with a duration of 5 1 2 /is. Using the video mixer featured elsewhere in this issue the sync pulses may be combined with picture information to give a composite video signal, in which case both circuits should be operated from a 5 V supply. The clock oscillator should be tuned to 1 25 kHz using a frequency counter, if available. Alternatively it may be adjusted to give a stable raster on a TV set. (see circuit 74) improved The 723 is a very widely used IC regulator. Hence the following cir- cuit, which is intended to reduce power dissipation when the 723 is used with an external transistor, should prove very popular. According to the manufacturer’s specification the supply voltage to the 723 should always be at least 8.5 V to ensure satisfactory operation of the internal 7.5 V reference and of the IC’s internal differential amplifier. Using the 723 in a low-voltage high- current supply, with an external series transistor operating from the same supply rail as the 723, invariably results in excessive dissipation in the series transistor. For example, in a 5 V 2 A supply for TTL about 3.5 V would be dropped acorss the series transistor and 7 W would be dissipated in it at full load current. Furthermore, the reservoir capacitor must be larger than necessary to prevent the supply to the 723 falling below 8.5 V in the ripple troughs. In fact the supply voltage to the series transistor need be no more than 0.5 V above the regulated output voltage, to allow for its saturation voltage. The solution is to use a separate 8.5 V supply for the 723 and a lower voltage supply for the external transistor. Rather than using separate transformer windings for the two supplies, the supply to the 723 is simply tapped off using a peak rectifier Dl/Cl . Since the 723 takes only a small current Cl will charge up to virtually the peak voltage from the bridge rectifier, 1.414 times the RMS voltage of the transformer minus the voltage drop of the bridge. The transformer voltage thus needs to be at least 7 V to give an 8.5 V 723 supply to the 723. However, by suitable choice of reservoir capacitor C2 the ripple on the main unregulated supply can be made such that the voltage falls to about 0.5 V above the regulated output voltage in the ripple troughs. The average voltage fed to the series transistor will thus be less than 8.5 V and the dissipation will be greatly reduced. The value of Cl is determined by the maximum base current that the 723 must supply to the series output transistor. As a rule of thumb allow about 10 /tF per mA. The base current can be found by dividing the maximum output current by the gain of the transistor. A suitable value for the main reservoir capacitor C2 is between 1500 and 2200 f/F per amp of output current. elektor july/august 1978 - 7-19 HF current gain ^)U tester The high-frequency current gain of a transistor is dependent on the DC bias conditions under which it operates, maximum gain being obtained at only one particular value of collector current. This simple circuit is designed to determine the optimum collector current for any NPN RF transistor. The transistor under test (TUT) is inserted into an amplifier stage which is fed with a constant amplitude 100 MHz signal from an oscillator built around T 1 . This signal is amplified by the TUT, rectified by D1 and filtered by RIO and C9 to give a DC signal proportional to the RF signal output from the TUT. This, in turn, is proportional to the gain of the TUT. The collector current through the TUT can be varied between 1 mA and 1 0 mA by means of PI , which should be fitted with a scale marked out linearly between these values. It is then a simple matter to adjust PI until the maximum output voltage is obtained on the meter, whereupon the optimum collector current can be read off from the scale of P 1 . This IR transmitter may be used with the receiver described elsewhere in this issue to make a simple infra-red control link. The IR signal is pulsed on and off at 24 kHz to enable the receiver to differentiate between it and extraneous ‘DC’ IR sources such as the sun and room lighting. To obtain a reasonable operating range the receiver must be sensitive and selective, and to avoid unreliability due to the transmitter frequency drifting outside the receiver passband the transmitter frequency must obviously be stable. To this end an LC oscillator circuit is employed, which is a transistor version of the Franklin oscillator commonly employed in valve circuits. Due to the Q of the tuned circuit the voltage across L 1 may exceed supply voltage. This could result in the collector-base junction of T1 being forward-biased, which would damp the tuned circuit. The inclusion of D1 prevents this occurrence. The oscillator frequency may be varied between approximately 23.7 kHz and 25.9 kHz using C3, which allows alignment of the transmitter and receiver. In view of the narrow bandwidth of the receiver and the limited tuning range available 7-20 - elektor july/august 1978 infra-red transmitter it is essential that the component values given in the circuit should be adhered to (use 5% tolerance components). Various types of infra-red LED may be used in the transmitter circuit, but the LD271 is most efficient and will give the greatest range. Whatever type of LED is used the performance of the circuit can be optimised by adjusting the value of R3 so that the LED current is 100 mA. Two or more LEDs may also be connected in series, in which case the value of R3 must also be adjusted to give a LED current of 100 mA. If more than two LEDs are connected in series then the supply voltage must be increased by 1.5 V for each additional LED. (see circuit 21) analogue-digital converter ^ ' Construction of an A/D converter is not a simple matter, since the circuit often requires the use of a number of precision components. However, the accuracy of the circuit described here is independent of component tolerances and is determined solely by the stability of a single reference voltage. 1C1 functions as a comparator. So long as the voltage on its inverting input is less than the analogue input voltage on its non-inverting input the output is high. FF1 receives pulses from the clock oscillator constructed around N1 and N2. Whilst its D input is held high by the output of IC1 its Q output remains high. CMOS switch SI is closed, while S2 is open, so C2 charges from the reference voltage (U r ef) via SI and R2. When the voltage on C2 equals that at the non-inverting input the output of IC 1 goes low. However, C2 continues to charge until the next clock pulse, when the Q output of FF1 goes low, SI opens and S2 closes. C2 now discharges through R2, R5 and S2 until the voltage on it falls below the analogue input voltage, when the output of IC 1 again goes high. On the next clock pulse the Q output of FF1 again goes high and the cycle repeats. Since C2 is charging and discharging exponentially it follows that the higher the analogue input voltage the longer will be the charge periods of C2 and the smaller will be the discharge periods. The result is that the output of IC1 is a squarewave whose duty-cycle is proportional to the analogue input voltage. Note that this only applies once the circuit has reached equilibrium. It does not apply during the initial phase when C2 is charging from zero. When the ‘start conversion’ switch is closed flip-flop FF2 is set. This enables counters IC5 and IC6. Both count clock pulses, but while IC6 counts every clock pulse IC5 counts clock pulses only whilst the Q output of FF 1 is high. When the Qi 2 output of IC6 goes high FF2 is reset and the conversion ceases. The count which IC5 has reached is thus proportional to the duty-cycle of the Q output of FF1 , which is proportional to the analogue input level. If the reference voltage is exactly 2.048 V then the count of IC5 will be 1000 for an input of 1 volt. The linearity of the prototype circuit was 1%, but this could probably be improved by using an LF 357 for IC1 , although a symmetrical supply will then be required. It is also possible to vary the clock frequency by changing the value of C3 (minimum 390 p for 50 kHz). To set up the circuit the input is grounded and PI is adjusted until the count from IC5 is zero. To check operation of the converter the reference voltage is connected to the analogue input when all outputs of IC5 should be high (count 2047). elektor july/august 1978 — 7-21 stereo width control Although the idea is not new, this circuit for a stereo width control is distinguished by its simplicity. A stereo width control is used to vary the width of a stereo sound image from mono, through normal stereo , to extended or super-stereo. Expansion of the stereo image width is obtained by means of negative crosstalk between the two channels, i.e. a portion of the L-signal appears in antiphase in the R-channel, and vice-versa. Positive crosstalk, where the ‘crosstalk’ signal portion and the channel into which it is blended are in phase with one another, results in a reduction of the stereo width. How the circuit functions is quite simple. Two opampsand resistors R2, R2' and R4 provide 60% (-4.4 dB) negative crosstalk at the outputs of IC1/IC1', whilst R3, R3' and PI provide variable positive crosstalk. With PI set for maximum resistance, the negative crosstalk at the outputs amounts to approx. 50% (—6 dB). With PI at the minimum setting (turned fully anti-clockwise), the L and R output signals are the same (mono), whilst with PI in the mid- position the negative and positive crosstalk cancel each other out, leaving normal stereo. Normal stereo can be obtained even more simply by incorporating a two-way switch, S 1 , in series with R4/P1 . hum filter using electronically simulated inductor 1 Qo-czb^-o* There are many cases where it is useful to be able to get rid of spurious mains (50 Hz) interference. The simplest way of doing this is to employ a special filter which rejects only the 50 Hz signal components whilst passing the other signal frequencies unaffected, i.e. a highly selective notch filter. A typical circuit for such a filter is given in figure 1. Since a filter with a notch frequency of 50 Hz and a Q of 10 would require an inductance of almost 150 Henrys, the most obvious solution is to synthesise the required inductance elec- tronically (see figure 2). The two opamps, together with 7-22 — elektor july/august 1978 ijo-t-b- -O* ©- — L R2 . . . R5, C2 and PI, provide an almost perfect simulation of a conventional wound inductor situated between pin 3 of IC1 and earth. The value of inductance thereby obtained is equal to the product of the values of R2. R3 and C2 (i.e. L = R2 x R3 x C2). For tuning purposes this value can be varied slightly be means of PI . If the circuit is correctly adjusted, the attenuation of 50 Hz signals is 45 to 50 dB. The circuit can be used as it stands as a hum rejection filter in harmonic distortion' meters or as a hum filter for TV sound signals. back and front doorbell A Some types of ‘ding-dong’ door chimes are designed to give different signals for back and front doors. However, the majority of doorbells are not, and this article describes a circuit that will allow an ordinary doorchime to produce two different signals, a ding-dong signal for the front door and a dong signal for the back door. A small gimmick is also incorporated to thwart impatient individuals who repeatedly press the bellpush. When the front door bellpush is pressed the ding-dong signal will sound once and is then inhibited for about five seconds. The dong signal, which is less strident, is allowed to sound a maximum of once every two seconds. The circuit operates as follows: When the front bellpush (SI) is pressed Cl charges rapidly through D2, RIO and the base emitter junctions of T3 and T4. These transistors are turned on briefly, which causes the striker of the chime to move rapidly across and back, thus producing the ding-dong chime. However, the chime cannot sound again until Cl has discharged through R1 and R2, which takes several seconds after the bellpush is released. Repeated pressing of the button has no effect. When the back door button (S2) is pressed the monostable comprising T1 and T2 is triggered, T1 turns on and T2 turns off. C4 now charges slowly via R8 and R9. T3 and T4 thus turn on slowly, pulling the striker across very slowly so that the ‘ding’ chime does not sound. When the monostable resets after about 1 'A seconds C4 discharges quickly through D3 and T2. T3 and T4 turn off and the striker of the chime flies back rapidly, thus producing the ‘dong’ sound. If illuminated bellpushes are used then R1 and R3 should be rated at between 10 and 33 ohm 2 W to suit the bellpush lamps. Otherwise any value between about 4k7 and 47 k is suitable. The orginal bell transformer can be used. The bridge rectifier should be capable of handling at least 1 A. voltage mirror Previous issues of Elektor have already discussed a number of different ways of using a transformer with only one secondary winding to obtain both a positive and a negative supply voltage. This design is a further contribution to the discussion. The circuit uses a second bridge rectifier (D 1 ... D4) which, via Cl and C2, is capacitively -coupled to the transformer. Since the resultant voltage is DC-isolated from the transformer, to which the other rectifier (D5 . . . D8) is connected, the positive terminal of C3 can be linked directly to the 0V rail to give a symmetrical ± supply. Since (because of C 1 and C2) C3 is charged from a higher impedance than C4, this capacitor should have a higher value than C4, otherwise the internal impedance and ripple voltage of the negative supply will differ significantly from it’s positive counterpart. The working voltages of the capacitors should at least equal the peak value of the transformer voltage. With the values given in the diagram the circuit will supply approx. 0.1 A for a transformer voltage of 1 5 V and a ripple voltage of 1 V. Naturally enough all the capacitance values can be increased by the same factor in order to reduce the ripple voltage. As far as the bridge rectifiers are concerned, these should be adequately rated to withstand the speedy rectifier Precision rectifiers which use a diode in the feedback loop of an operational amplifier are well known. Such an arrangement virtually eliminates the forward voltage drop of the diode and allows even small signals to be accurately rectified. However, since the op-amp operates open loop up to the point where the diode becomes forward biased the maximum operating frequency of such rectifiers is limited by the slew rate of the op-amp used. Precise rectification of small signals even at the higher audio frequencies requires an op-amp with quite a high slew rate and such op-amps are not inexpensive. Fortunately an alternative solution is to build a precision rectifier using inexpensive small-signal transistors. In this circuit diodes D1 and D2 are current driven, so the output voltages developed across load resistors RIO and R1 1 are proportional to the current through the diodes and are independent of their forward voltage drops. The signal to be rectified is fed to T 1 , and current drive to the diodes is achieved by bootstrapping the emitter of T2 to the junction of R1 and R2. A positive half-wave rectified version of the input signal is available across RIO and a negative half-wave rectified signal at R1 1. When viewed on an oscilloscope there was no deviation from a true half-wave rectified output at frequencies in excess of 400 kHz and input signal levels up to 2 V peak to peak (sinewave input). The only setting up that the circuit requires is to adjust PI , with no input signal, until the collector voltage of T1 is exactly zero. in a variety of alarm circuits with resistive transducers such as LDRs and thermistors or, as in this case, a liquid level probe. When the electrodes of the probe are immersed in the liquid, which must, of course, be a conductor, an AC 7-24 - elektor july /august 1978 liquid. This is detected by a comparator in the IC, and an alarm signal is fed to the loudspeaker. The best material to use for the electrodes is stainless steel, since this is resistant to corrosion. The probe can easily be constructed from a pair of stainless steel meat skewers, which are available from hardware shops. (National Semiconductor Applications) FET audio amplifier Power FETs have been used in a number of Japanese audio amplifiers for some time now, and indeed were discussed in Elektor No. 14, June 1976, p. 628. Readers are referred to this article for a full discussion of the application of power FETs in audio amplifiers. Using power V-FETs manufactured by Siliconix it is now possible to present a FET audio amplifier design suitable for home construction, which is based on a Siliconix application note. The advantages offered by V-FETs in audio output stages are considerable. The 2N6658 used in this circuit has a cutoff frequency of 600 MHz, and yet is completely free from the secondary breakdown problems that bedevil high-frequency bipolar transistors. The current gain of a V-FET is virtually infinite, the transfer characteristic is extremely linear for drain currents greater than 400 mA and the temperature coefficient of drain current is negative, thus eliminating thermal runaway problems. The maximum drain source voltage of the 2N6658 is 90 V, which is more than adequate for audio amplifier applications. However, the maximum drain current is only 2 A and the maximum dissipation 25 W, so a number of V-FETs must be connected in parallel in the output stage of the amplifier (T8 to T1 3). The same type (polarity) of FET is used in each halve of the output stage, and the two halves of the output stage therefore require antiphase drive signals. This is easily achieved, as antiphase signals are available as far back in the circuit as the input stage, which consists of a long-tailed pair T1/T2. The antiphase signals from the collectors of the input stage drive a second long- tailed pair T3/T4, the antiphase outputs of which feed two driver stages, T5/T14/T15 and T6/T16/T17,each of which comprises a current mirror and cascode stage. Provision of DC biasing throughout the amplifier is simplified by the use of constant current (Norton) diodes. It should be noted that, although any type of Norton diode from CR390 to CR470 may be used for D3, D6 and D7, they must all be the same type. With the power supply shown, which is adequate for a stereo version of the amplifier, the circuit will deliver an output of 40 W per channel into 8 ohms with a harmonic distortion of 0.04% at 1 kHz. Clipping does not occur until 55 W into 8 ohms, but above 40 W the distortion will gradually increase. The slew rate of the amplifier is 100 V/jis, and the output is short-circuit proof. Finally, a few practical hints on constructing and setting up the circuit. The six output FETs should be mounted together on a single heatsink with a thermal resistance of less than 2°C/W. The gate resistors R15 to R22 should be mounted as close as possible to the gate leads of the FETs. For setting up the amplifier, it should temporarily be connected to a stabilised power supply with the current limit set to between 500 mA and 1 A. Alternatively a 100 ohm 10W resistor may temporarily be connected in series with the drain lead of T8 . . . T1 0 and of Til ... T1 3 to limit the current. Before applying power PI and P2 should be set to maximum resistance and a milliameter conrtected in the positive supply lead. When power is applied the current consumption should be about 40 mA. P2 should then be adjusted until the supply current shows a sharp increase, after which the amplifier should be left for about 5 minutes to warm up. The supply current may then be adjusted to between 200 and 350 mA using P2. Finally, PI should be adjusted to give minimum distortion at an output power of 1 0 W into 8 ohms with a 1 kHz sinewave input. However, if equipment is not available to carry out this adjustment PI may simply be set to its mid- position or adjusted by ear. Literature. Siliconix Application Note AN 76-3 and Design Aid DA 76-1. elektor july /august 1978 — 7-25 power flasher Despite the vast array of solid-state devices now available, the flasher units for car direction indicators are still almost exclusively electro- mechanical. Apart from the obvious objection of unreliability, these units suffer from the problem that the flashing rate is dependent on ambient temperature, battery voltage and load. This latter factor means that if one wishes to wire all four indicators to flash simultaneously as a hazard warning, it is necessary to use a separate flasher unit. The electronic flasher discussed here suffers from none of these disadvan- tages. The repetition rate is practi- cally independent of battery voltage, temperature and load, has a built-in hazard warning switch and is extremely reliable. Furthermore it complies with all the legal require- ments for turn indicators, the repetition rate of 40 to 90 flashes per minute being within the specified range and the circuit being arranged so that the indicators light immedi- ately when the turn indicator switch is operated. The circuit is basically an astable multivibrator constructed around two CMOS NOR gates N1 and N2. N3, N4, T1 , T2 and T3 buffer the output of this astable to drive the indicator lamps. When the indicator switch is oper- ated C2 discharges rapidly through D 1 and the indicator lamps. Pin 1 3 of N 1 goes high and its output goes low. The outputs of N3 and N4 thus go high, turning on Tl, T2 and T3 and lighting the indicators. The astable then begins to oscillate at approximately 1 Hz, turning the indicator lamps on and off. If the hazard warning switch, S 1 , is closed then the circuit operates in exactly the same fashion except that all four lamps are connected in parallel and flash in synchronism. T3, which switches most of the load current, must be mounted on a heat- sink. If a metal box is used to house the unit then T3 can be bolted to the wall of this using an insulating washer and bush. The current in the leads connected to points A and B is quite large (up to 8 A) so heavy- gauge wire must be used for these connections. The positive supply lead must be fitted with a 10 A fuse if not already fused. Parts list. Resistors: R1,R3,R4= 2M2 R2 = 100 k R5 = 4k7 R6 = 1 20 O (1 Watt) Capacitors: Cl - IOu/16 V C2 = 1 p/16 V (tantalum) C3 = 1 n C4 = 220 n Semiconductors: IC1 =4001 (B) Tl = BC 557, BC 177 T2 = BC 328, BC 327 T3 = FT 2955 (Fairchild) TIP 2955 D1 = 1N4148 7-26 — elektor july/august 1978 / signal injector This simple signal injector should find many uses in trouble shooting and alignment applications. It produces an output with a fundamental frequency of 100 kHz and harmonics extending up to 200 MHz and has an output impedance of 50 ohms. N1 , N2 and N3 form an astable multivibrator with a virtually sym- metrical squarewave output and a frequency of approximately 100 kHz. The output of the oscillator is buffered by a fourth NAND gate N4. + Since the squarewave is symmetrical it contains only odd harmonics of the fundamental frequency, the higher harmonics being fairly weak due to the relatively slow rise time of the CMOS devices employed. As it is necessary for the higher harmonics to be at a reasonable level if the circuit is to prove useful at high frequencies, the output of N4 is fed to a differ- entiating network R2/C2. This attenuates the fundamental relative T1,T2=TUN N1 N4 = ICls4011 to the harmonics, producing a needle-pulse waveform which is then amplified by T1 and T2. This is rich in harmonics and, due to the extremely small duty-cycle of the waveform, the power consumed by the output stage, T2, is fairly small. The output frequency of the signal injector can be adjusted by means of PI . If an accurate output frequency is required then the signal injector can be adjusted by beating its second flasher bleeper I Although extremely useful, the self- V cancelling devices fitted to car direction indicators are not infallible. For example, they will not operate when only a small movement of the steering wheel is made, as when pulling out to overtake. An audible warning device to indicate that the flashers have not cancelled is preferable to the visual indication normally fitted, as it is more notice- able and does not require the driver to take his eyes off the road. The circuit consists simply of a 555 timer connected as a 1 kHz astable multivibrator. The output of the 555 is more than sufficient to drive a small loudspeaker. When the trafficator switch (S) is set to either the left or right positions, power is supplied to the multi- vibrator via the flasher unit and D1 or D2. The circuit thus ‘beeps’ with the same rhythm as the flashing of the indicators. If desired, the volume can be reduced by increasing rhe value of R3. The ‘beep’ harmonic with the 200 kHz Droitwich broadcast transmitter. The frequency stability of the circuit is largely determined by the construction. To minimise hand capacitance effects the unit should be housed in a metal box for screening, with the only output connection being the signal probe. If desired, a 1 k preset may be included in series with P 1 to allow easier fine tuning. frequency is determined by Cl . For operation in positive earth cars D1 and D2 should be reversed and the multivibrator circuit turned upside-down, i.e. the Cl /pin 1/ loudspeaker junction is connected to the commoned anodes of the diodes and the Rl/pin 4/pin 8 junction is connected to supply common. elektor july /august 1978 — 7-27 ultrasonic alarm receiver 41 LT This ultrasonic receiver can be used with the transmitter described elsewhere in this issue to construct an alarm system operating on the Doppler principle. Ultrasonic signals sent out by the transmitter are reflected from objects in the area under surveillance and are picked up by the receiver. Reflections from any moving object such as an intruder will exhibit a slight shift in frequency due to the Doppler effect. Mixing of the Doppler shifted signals with the normal reflections causes cyclic variations in the amplitude of the received signal at a low frequency, dependent on the speed of the moving object. These variations are detected by the receiver circuit and used to trigger the alarm. The receiver utilises the reflex principle. Ultrasonic signals picked up by the receiver transducer are amplified by T1 and T2. A tuned circuit LI /Cl , connected across the transducer, improves selectivity. Due to a lowpass filter R9/C7, the amplified U/S signal cannot reach the base of the low-frequency amplifier stage, T4. Instead it passes through C5 to be rectified by an ‘infinite impedance' detector stage built around FET T3. A lowpass filter comprising R7 and C2 removes the high frequency component of the signal, whilst C3 acts as a DC blocking capacitor. The signal which appears across C2 is thus the low-frequency envelope of the received ultrasonic signal, which of course results from variations in the received signal amplitude due to Doppler shift. The LF signal passes through LI , which is virtually a short circuit at low frequencies and passes through T1 and T2. These transistors are thus used to amplify both U/S and LF signals, as in a reflex radio receiver. After amplification the LF signal passes through the lowpass filter R9/C7 to the output stage T4/T5. Depending on the setting of P2 this stage can operate as a Schmitt trigger or a linear amplifier. In the trigger mode T5 is normally turned on so that the output is high. When a signal arrives at the base of T4 the output will go low. In the linear mode the LF signal can be made audible by a pair of headphones or a small loud- speaker connected to the output. To set up the receiver P2 is adjusted (with no U/S input Signal) until the output goes high. The transmitter is then switched on and PI is adjusted to give the required sensitivity. (see circuits 44 and 4 7) debouncer Various microprocessor circuits place particular requirements upon the duration of certain control signals. If these signals are generated by manually operating a switch or key (e.g. reset and interrupt keys), then the conventional debounce circuit consisting of an RS flip-flop is not always foolproof, since there is always the chance that the key will be released prematurely. The accompanying circuit however, ensures that the signal level is maintained for a certain time after the key has been released. The exact length of time is determined by the values of R I , R2 and Cl for switch 5 1 , and by R3 , R4 and C2 for switch 52. If SI is depressed, then output 1 goes low, whilst depressing S2 will likewise take output 2 low. Since the 556 has open collector outputs, it can easily be connected in a wired-OR configuration. elektor july /august 1978 — 7-29 ultrasonic alarm indicator This circuit can be used to link one or more ultrasonic receivers (see circuits 44 and 45) to a central alarm system. It can also be used in conjunction with the infra-red alarm described elsewhere in this issue. The circuit provides both audible and visual (LED) indication that the alarm has been triggered. The indicator consists of a number (in the circuit shown there are three) of flip-flops, each of which is connected to the output of a receiver. The flip-flops are derived from the ‘super-simple touch switch’ which is also described elsewhere in this issue. The flip-flop is triggered by a logic ‘0’ appearing at its input (the output of the receiver of course goes low when the alarm is activated). The correspond- ing LED then lights up and remains on until the flip-flop is returned to its original state by pressing the reset button. One or more of the inputs going low also has the effect of triggering the monostable round N4. It in turn drives the squarewave generator round T1 and T2. The resulting squarewave signal is fed via an output stage, T3, to the loudspeaker. By means of PI the duration of the alarm signal can be varied from one or two to several tens of seconds: volume control is provided by P2. The number of inputs can be increased indefinitely by simply repeating the circuit around the input flip-flop the desired number of times. Using one 4050 an indicator circuit for five alarm installations can be built. (o) selective bandpass @1 filter There are a number of applications, such as the analyser filters in a real- time audio spectrum analyser, which require bandpass filters that are -30 — elektor july /august 1978 highly selective and yet have a virtually flat response within the passband. Simple selective (resonant) filters fail to meet these requirements because selectivity requires a high Q (quality factor) whereas a flat response within the passband re- I quires a low Q. These conflicting requirements cannot be satisfied by . single filter. A solution is to connect in cascade two selective filters with staggered centre frequencies. Each filter has the same gain (A) at the centre frequency and quality factor Q, but the centre frequencies are different (f G i and fo 2 )• The centre frequency of the cascaded combination is f 0 , the frequency at which the two response curves intersect. By ensur- ing that the gains of the individual filters are at this intersection, the combined gain at fo is A and the response is maximally flat within the passband. A practical circuit for such a filter arrangment is shown in figure 2. Given the desired centre frequency f Q and either the required Q or bandwidth B, the component values R1 , R6 and Cl to C4 can be cal- culated using the equations given. Stable-Start-Stop- Squarewave v In digital circuits where parallel information must be converted into serial data, a start-stop oscillator is often used. One system is to use the oscillator to clock a counter, the output of which is compared with the parallel data. Initially the counter is reset; the external oscillator is then started, clocking the counter; when the correct count is reached, the oscillator is stopped. The result is a (clock) pulse train, the length of which corresponds to the binary number given by the parallel data. It is not sufficient for these applications to gate the output of a free-running oscillator, since the ‘enable’ signal is not normally synchronised to the oscillator. The circuit described here is actually turned on and off by the enable signal, and it has proved reliable and stable for output frequencies up to 10 MHz. As long as the enable input (one input of N3) is at logic 0, the oscillator is blocked and the output of N4 is also held at logic 0. When the enable input becomes logic 1 , the oscillator starts immediately and the first output pulse is delayed only by the propagation times of N3 and N4. logic 0= STOP 78103-1 0— r I nrirL JrLn_ruL_ elektor july/august 1978 - 7-31 r Eif^i bicycle speedometer The novel feature of this bicycle speedometer is that it switches itself on when the bicycle starts to move and switches off when the bicycle stops, thus prolonging battery life without the need for a manual on/off switch. Speed sensing is carried out by a reed switch attached to the bicycle frame, which is actuated by a magnet or magnets fixed to the wheel spokes. Electronics purists who scoff at ‘unreliable’ electromechanical switches need have no fears for the long-term serviceability of this arrangement. The life of a reed switch is typically 1 0 8 operations. Even with a small ( 1 0 inch) wheel bicycle this gives a life of 10® x 10 x 7r inches or 49,583 miles! The circuits operates as follows: When the bicycle is stationary T2 is turned off, C2 is charged to +9 V via R2, so T1 is turned off and no power is supplied to the circuit. As the bicycle starts to move the changeover reed switch S 1 switches between positions B and C, thus turning T2 on and off. Since the reed switch is AC coupled to T2 no current is drawn when the bicycle is at rest, even if the reed switch should be activated by the magnet in this condition. When T2 is turned on C2 discharges rapidly through D2, turning on T1 and supplying power to the circuit. After the bicycle stops it takes several seconds for C2 to recharge sufficiently forTl to turn off. T2 also triggers IC1 , which is connected as a monostable multivibrator. The output pulse width is fixed, so as the speed, and hence the triggering frequency, increases, the duty-cycle of the output waveform becomes greater. The average output voltage, which is measured by the meter, thus increases in proportion to the speed. 7-32 - elektor july/august 1978 f To calibrate the circuit a small calculation is necessary. The input frequency for a given speed is obtained from the equation: where ‘n’ is the number of magnets used, ‘s’ is the speed in miles per hour and ‘D’ is the wheel diameter. The input frequency for a given speed can thus be calculated, and the speedometer can be calibrated by feeding in this frequency from an audio oscillator and adjusting PI until the correct speed reading is obtained. As an example, suppose a bicyle with a 10 inch wheel is being used, and the meter is to be calibrated to a maximum speed of SO m.p.h. used on the wheel to bring the equivalent speed down to a more reasonable 44.5 m.p.h. For a 20 inch wheel the situation is worse as this wheel turns at half the rate of a 1 0 inch wheel for a given speed, so four magnets would be required. Since SPDT reed switches are some- what rare, figure 2 shows how to connect two single pole reed switches and their placement on the wheel so they will function like S 1 in figure 1 . As with almost all self-driving ‘melody generators’, this circuit con- sists of a current-controlled oscillator and a control system. One possible approach would be to use a PLL [ (phase locked loop). A variant of the PLL concept is the sample-and-hold PLL, that has the advantage of being frequency sensitive to a sawtooth voltage. The ‘sawsong’ is therefore a sample-and-hold PLL that is prevented, by a deliberate wrong- polarity connection of the oscillator signal, from acquiring lock. Figure 1 gives the block diagram and figure 2 the complete circuit diagram. The current-controlled oscillator con- sists of current-mirror T 1 + T2 and ► unijunction transistor T3. This multi- vibrator generates a sawtooth wave- form that is buffered by T4. T5 + T7 form a well-known sample-and-hold Assuming one magnet is used then the frequency for 50 m.p.h. is 1 x 50x 528 30 x 3.142 x 10 i.e. 28 Hz. An alternative is to feed in a 50 Hz signal from the low voltage secondary of a mains transformer (6-12 V). The equivalent speed can be calculated by re-arranging the previous equation. _ 30 x 7T x D x f S 528 xn However, in the previous example 50 Hz would correspond to a speed of about 89 m.p.h. so if 50 Hz is to be used as a calibration frequency, two magnets should be saw-song circuit, for which the sampling pulses sound of a singing saw. With S2 open are generated by another unijunction and S 1 closed the sound is somewhat transistor T8 and passed via T6 to the gate of T5. When the circuit is operated with ‘jumpier’. With both switches closed the circuit will produce a parakeet- like squawking noise. Potentiometer switch SI open and S2 closed, it will PI can be used to vary the tempo of produce a signal resembling the the ‘melody patterns’. T1, T2 = BC557A.E T3,T8 = 2N2646 T5, T7 = BF 256A elektor july /august 1978 - 7-33 preconsonant As explained in greater detail in the Consonant article, the design aim for both units was to achieve superior performance in a true home-construc- tion project. For this reason, readily available transistors have been used throughout. It may not seem ‘up-to-date’ if a circuit doesn’t contain at least one IC, but the fact is simply that integrated circuits are either not good enough or else not readily available to the home constructor. A 741, for instance, makes for a notoriously sub-standard audio design, whereas a TDA 1034N (which does offer good performance) is any- thing but readily available — in fact, it is doubtful whether many of our readers had even heard of it before reading this issue! On the other hand, the BC 109C and its relatives (BC 249C, BC 549C etc., see the TUP/TUN list elsewhere in this issue) is available all over Europe — and, for those of our American and Canadian readers who have remarked that a code number like BC 109 reminds them of a jet aircraft, almost any high- quality low-noise Silicon transistor will perform equally well in these circuits. The circuit To some of our readers, the circuit (figure 1) may seem vaguely familiar: it is derived from the popular Preco design. With good reason: over the years it has proved its reliability, and every- one who has measured its performance has been astounded. Four transistors per channel may seem excessive, but time and theory can prove that two transis- tors are not really sufficient; three transistors are possible; but four transis- tors make for a reliable circuit - and the choice makes very little difference in the total cost of the project. The input impedance is determined by R 1 , R3 and the input impedance of T 1 . With the values shown, it is very close to the required 47 k. The signal-to-noise ratio of the preamp is mainly deter- mined by Tl. A high S/N ratio can be obtained by choosing a good transistor, setting the collector current and the collector-emitter voltage correctly, and selecting the optimum emitter im- pedance. The base impedance is also important, but for a disc preamp there is very little lee-way here since this 7-34 — elektor july /august 1978 The Consonant, described elsewhere in this issue, is a high- quality audio control amplifier. Its main features are: extremely good performance, use of readily available components and ease of construction. The Preconsonant described in this article is a matching disc preamp. It can be mounted on the Consonant p.c. board, it too uses readily available components, and its performance is exceptionally good. Although the printed circuit board is designed to match the Consonant board, the disc preamp is a self-contained unit that can be used in conjunction with any high- quality control amplifier. Specifications: maximum deviation from IEC (RIAA) frequency response input overload level @ 1 kHz: input headroom: signal-to-noise ratio: dynamic range: distortion at +14 dB > 200 mV (RMS) > 32 dB* >72 dB* > 100 dB ’ input headroom, signal-to-noise ratio and reference level for distortion measurement are referred to 0 dB = 5 mV (RMS) input level at 1 kHz; see text. impedance is determined by the re- quired input impedance (47 k) and the I impedance of the phono cartridge. For the type of transistor specified (and I for most other low-noise, silicon PNP 1 transistors) the fixed base impedance^ corresponds to minimum noise at a I collector current of approximately I 100 pA. Tl drives a simulated super- I NPN transistor, consisting of T2 and T3. The collector impedance of the ‘super-transistor’ consists of a current source, T4. This particular configuration combines several attractive features: high gain, good supply ripple rejection I (as will be explained later) and high I current drive capability. The latter I feature is important because this stage I must also drive the feedback networks (R1 1 , C5 . . . C7), and since the’] feedback network must be tailored to I provide the desired IEC correction* its I impedance drops sharply with increasing I frequency. Theoretically, the IEC frequency re- I sponse curve corresponds to three time I constants: 3180/us(50Hz, pole),318/is I (500 Hz, zero) and 75 /is (2120 Hz, I pole). In practice, in a combined circuit I like this, mutual interaction of the I various RC networks calls for slightly I different time constants. R11,C5, C6^ and C7 provide the two higher timeT* constants; the lower is determined in > part by R6, R7, R8 and C4. Back now, briefly, to T4. Basically a current source consists of one transistor I and three resistors: T4, R13, R14 and R15. However, including C9 ensures | that the base voltage of T4 is identical to the voltage at the lower end of R15, I for all but the very lowest frequencies 1 (and DC). This ensures that the AC I collector current is virtually zero, I * In Europe, the frequency characteristic of disc preamps must conform to the IEC norm. The (old) RIAA norm specified in the USA basically specifies the same characteristic in a different way; the new RIAA norm is vir- tually identical to the combined Preconsonant/ Consonant characteristic since it specifies an additional low-frequency roll-off similar to that incorporated in the Consonant. providing a very high AC impedance for T2/T3 and simultaneously leading to a very high supply ripple rejection. Remember the ‘One-TUN gyrator’? The printed circuit board The complete stereo disc preamp can be mounted on the p.c. board shown in figure 2. As stated earlier, this board was designed for mounting on the Consonant board: the six solder pads along one side of the board correspond to six pads on the Consonant board, so the two boards can be joined by means of six short wire links. There are four true electrical connec- tions (left and right preamp outputs, • supply common and positive supply); the remaining two connections are included for mechanical rigidity. The positive supply voltage (21 V) is derived from the Consonant supply. If the Preconsonant is to be used in combi- nation with a different control amplifier, a suitable supply (20 ... 24 V, 10 mA) will have to be derived from this. The only other connections to the Preconsonant are the left and right inputs; these are connected by means of short lengths of screened cable to the '••disc preamp input. These connections are illustrated in greater detail in the Consonant article (figure 10). Preset potentiometer PI is included in case the output of the Preconsonant is too high. The ‘Tuner’ and ‘Auxiliary’ inputs of the Consonant are also fitted with input level presets, so that it is always possible to reduce the level of the two loudest signal sources to that of the third. In this way, annoying level jumps when switching from one signal source to the next can be avoided. If the Preconsonant is used in conjunction with some other control amplifier, PI can usually be set at maximum. Performance The main specifications are given in a separate table. The frequency response and distortion specifications are also illustrated in the graphs shown in figures 3 and 4. Figure 3 is the frequency response as plotted on a Briiel & Kjaer recorder. The Figure 1. Complete circuit for one channel of the Preconsonant. required levels in dB (with respect to 0 dB at 1 kHz) are listed below the frequency scale. The deviation from the required response is also plotted, along the 0 dB line of the (relative) dB scale. As can be seen, the prototype gave the required response curve with a deviation of considerably less than ± 0.5 dB over the complete 20 Hz ... 20 kHz fre- quency range. However, owing to the 5% tolerance in the frequency-deter- mining components, a deviation of up to ± 1 dB can theoretically occur (“worst case’). elektor july/august 1978 — 7-35 Figure 4 shows the result of a difference- miss. However, the Preconsonant stood frequency distortion measurement at an up well to this test: the average input level of + 15 dB (30 mV at 1 kHz), measured distortion was around 0.01%, using an ‘anti-RIAA network’ to main- with the highest peaks still well below tain this level over the entire frequency the 0.03% level. Note that this measure- range. A difference-frequency distortion ment was performed at the highest measurement uses two equal-amplitude input level likely to be encountered! sinewaves, fi and f 2 , with a constant This point may require some further frequency difference f 2 - fi as the two clarification. Throughout this article, signals are swept through the audio 0 dB has been consistently taken as band. This arrangement has the advan- corresponding to 5 mV (RMS) at 1 kHz. tage that measuring the difference This value is derived as follows. On frequency component at the output of a modern records, 0 dB level (correspond- (pre-)amplifier under test gives infor- ing roughly to the average level in loud mation, not only concerning the (steady- passages) is typically about 4 cm/s peak state) Total Harmonic Distortion, but velocity at 1 kHz. Modem hifi cartridges, also on several other ‘nasties’ which a by and large, deliver approximately ‘normal’ distortion measurement would 0.5 ... 2 mV (RMS) per cm/s tip peak Figure 3. Frequency response and deviation from the required response as measured for the prototype. Figure 4. Result of a difference-frequency distortion measurement at an input level of +15 dB (30 mV at 1 kHz). 7-36 — elektor july/august 1978 velocity. This means that 0 dB on the adequate insurance for technological 0 dB = 4 cm/s), so even in this extreme record may correspond to 2 ... 8 mV at improvements for some time to come! case (with the least sensitive cartridge) the input of the disc preamp, at 1 kHz. On the other hand, to be completely the Preconsonant is still a good 10 dB 5 mV is reasonable intermediate value, safe the signal-to-noise ratio should be better. Good enough! H However, this is not the full story. On at least 6 dB better than that of a modern records, instantaneous signal modern record, even with the least sen- peaks of up to +14 dB may occur, and sitive cartridge (approximately 700 )iV for this reason the distortion measure- per cm/s peak velocity). 0 dB in this ment was carried out at approximately case corresponds to 2.8 mV (RMS), or this level. For complete comfort, the about 6 dB less than the value assumed preamp should be able to handle a so far. Since the Preconsonant has a + 14 dB instantaneous peak with the signal-to-noise ratio better than 72 dB most sensitive of modern cartridges - with respect to the 5 mV r 'erence corresponding to some 60 mV (RMS) at level, the S/N ratio even with tl. 'east 1 kHz. The input overload level of the sensitive cartridge will be over 6 j dB. Preconsonant is well over 200 mV, Manufacturers estimate that the best providing a safety margin of a good S/N ratio possible with a first-rate LP 10 dB above even this extreme level — pressing is about 56 dB (with respect to elektor July /august 1978 — 7-37 INSONANT * CONSONANT - CONS( ] Specifications (see also text) gain max. output voltage nominal output voltage signal-to-noise ratio overload margin total harmonic distortion channel separation dynamic range output noise level tone control characteristics: bass: turnover point 150 Hz turnover point 300 Hz treble: turnover point 2 kHz turnover point 4 kHz filters: rumble filter: turnover frequency scratch filter: turnover frequency balance control range current consumption (without disc preamp and PPM) 20 Hz ... 50 kHz (+0 dB. -3 dB) 1 47 mV RMS for 440 mV RMS out x 3 (9.5 dB) 3.5 V RMS (lOVpp) 440 mV RMS > 72 dB for 440 mV RMS out > 15dB above 440 mV RMS out approx. 0.04% (for 440 mV out) > 50 dB (at 1 kHz) > 90 dB approx. 0.1 mV RMS t 8 dB (at 50 Hz) 10 dB (at 50 Hz) 12 dB (at 10 kHz) 8 dB (at 10 kHz) 60 Hz (-3 dB), 1 2 dB/octave 10 kHz (—3 dB), 12 dB/octave +2 dB, -7 dB approx. 30 mA (including LED D2I The Consonant is a high-quality audio control amplifier designed to complement the best modern power amplifiers. It offers such refinements as scratch and rumble filters, tone controls with cancel facility and switchable turnover frequencies, and provision for a built-in LED signal level meter. All components, including potentiometers and switches, are mounted on a single p.c. board, thus greatly simplifying wiring. A compatible disc preamp, which may be mounted separately or fixed to the back of the main board, is described in a separate article ('Preconsonant'). (Designed in cooperation with T. Meyrick) 7-38 — elektor july/august 1978 The principal considerations which governed the design of the Consonant were that: 1. The performance and facilities offered should be comparable with those provided by the best commercial designs. 2. The circuit should be simple to con- struct and should use readily-available components. 3. The controls should be laid out in a clear and logical fashion for ease of operation. Block diagram Figure 1 shows a block diagram of one channel of the Consonant, the layout of which follows conventional practice for a control amplifier of this type. Any one of three signal sources, disc, tuner or auxiliary, may be selected by the input switch, and the input sensi- tivities may be adjusted by means of a preset on each input (except disc). Immediately after the input selector comes the rumble filter. This is placed before the tape monitor switch and can i INANT * CONSONANT • CONSONA] thus be switched in when recoiding from disc onto tape. There would be j» f little point in putting the rumble filter after the tape monitor switch, for the I simple reason that tape recorders do not I suffer from rumble. Any rumble present I on a disc that is to be recorded should I be filtered out before the signal is fed to I the tape deck. I At the output of the rumble filter the I signal from the source is available for I feeding out to the tape recorder and the I tape monitor switch allows either this signal, or the signal fed back from the tape recorder, to be routed through the "^control amplifier. This is extremely * useful if the tape recorder is a three- head machine, as the signal being I monitored is the actual signal that has I been recorded on a tape. If the recorder I is a two-head machine then it merely . feeds back the original signal. I The scratch filter is placed after the tape I monitor switch since the facility to suppress high-frequency tape noise is I just as useful as the facility for sup- I pressing record surface noise. In I addition, since many cassette recorders have a fairly limited h.f. response to begin with, switching in the scratch I filter in the record path could result in an extreme lack of treble while leaving the tape noise unaffected. If a noisy I record is to be transcribed onto tape I it is much better to switch in the scratch filter during playback, thus attenuating the recorded disc noise and the tape noise. Baxandall-type bass and treble controls are incorporated into the output stages of the control amplifier, and a stereo image width control is also provided which can vary the image width from mono through stereo to ‘super stereo’. Both the tone controls and image width control are provided with cancel switches. Finally, at the output of the control amplifier is the channel balance control. Of course, since the sensitivity of each input can be individually adjusted, the balance control will rarely be used once the system is set up, except to compensate for unbalanced programme material. Complete circuit The complete circuit of one channel of the Consonant is given in figure 2 and observant readers will notice that the tone control section bears a marked similarity to that of the highly success- ful Preco preamp, which was published in Elektor Nos. 1 2 and 13. However, the input section of the Consonant is con- siderably more complex than the Preco because of the rumble filter, scratch filter and the tape monitor facility, which the Preco lacked. The input signals, with the exception of the signal from the disc preamp, arrive first at presets PI and P2, which are used to set the input sensitivity for tuner and auxiliary inputs. The desired signal is then routed to the first stage of the Consonant via input selector switch SI. T1 is connected as an emitter fol- lower and acts as a buffer between the signal sources and the rest of the circuit. This arrangement is superior to feeding signals direct to the tape monitor switch, as is the case with some ampli- fiers, since the change in load impedance when the monitor switch is operated can cause variations in signal level if the signal source is unbuffered. The rumble filter is also constructed around Tl. It has a slope of 12dB/ octave and a turnover frequency (-3 dB point) of 60 Hz. As is evident from the circuit diagram and figure 4, the rumble filter is not switched com- pletely out of circuit even when the filter switch is in the ‘off’ position - the turnover frequency is simply moved down to 20 Hz. This ensures that sub- sonic frequencies caused by record warps, or lowering of the pickup onto a record, are severely attenuated. This is a desirable feature since these low frequencies, although inaudible, can damage the bass units of loudspeakers. The output of the rumble filter is fed via the tape monitor switch to the gain control P4. The signal from the gain control is buffered by a second e.mitter follower, T2, around which the scratch filter is built. The slope of the scratch filter is 12 dB/octave, and the turnover frequency is 1 0 kHz. Like the rumble filter, the scratch filter is never switched completely out of circuit, but the turn- over point is shifted up to 50 kHz with S5 in the ‘off’ position. This prevents the Consonant from operating as a long-wave radio receiver, which can happen in preamps whose frequency response is not restricted in this way. The slew-rate of signals with fast rise times is also limited, which helps prevent the power amplifier from running into transient intermodulation elektor july/august 1978 — 7-39 distortion (TIM). Of course, an upper frequency limit of 50 kHz is more than adequate for high-fidelity reproduction. The to.ne control circuit is very similar to that employed in the Preco, the prin- cipal differences being the cancel switch S8, which shorts out the frequency- selective networks around P5 and P6, and the turnover point selection switches S6 and S7, which allow additional capacitors, C14 and Cl 6, to be switched into circuit. It will be noted that the rumble and scratch filter switches and the turnover point switches are all shunted by high value resistors. These ensure that the capacitors they control have the same DC voltage applied to them whether they are switched into circuit or not. This eliminates the clicks that would otherwise occur due to the capacitors charging when switched into circuit. The overall gain of the Consonant (x3) is provided by the transistors in the tone control section, T3 and T4. At the output of the Consonant is the balance control, P7. R33 and P7 are AC coupled to the collector of T4 via C20 and thus form part of the collector load resistance which deter- mines the gain of this stage. As P7 is turned anticlockwise its resistance increases, increasing the collector load resistance of T4 and hence the gain of the left channel. P7', the balance con- trol for the right channel, is connected the opposite way rount to P7', so that as the control is turned anticlockwise its resistance falls and the gain of the right channel decreases. When the balance control is turned clockwise the resistance of P7' increases while that of P7 decreases, so the gain of the right channel rises whilst that of the left channel falls. With the balance control central the resistance of P7 equals that of P7', so the gain of both channels is, of course, the same. This arrangement has advantages over the single-gang potentiometer which performed a similar function in the Preco. Due to the contact resistance between wiper and track not being zero, crosstalk could occur along the balance potentiometer track in the Preco. With the two-gang potentiometer used in the Consonant this cannot occur. However, a single-gang poten- tiometer may be used if desired, as will be described later. The final control in the Consonant is the stereo image width control. Various explanations of the operation of width controls have been given in Elektor, so only a brief description will be given here. When S4 is closed the left and right channels are linked via R35 and P3. R35 joins the emitters of T4 and T4' and thus effectively converts these two stages into a differential amplifier. The signal that now appears at the collector of T4 will thus no longer be simply the left channel signal, but L -kR, where k is a constant determined by the circuit parameters. In other, words the signal in the left channel will consist of the original left signal plus an antiphase contribution from the right channel, this being the significance of the minus sign. Similarly, the right channel signal will consist of R - kL. The effect of an antiphase right channel in the left channel is to make the right channel signal appear even further to the right, and a similar effect is apparent in the left channel, i.e. the image width is greater than normal. P3, on the other hand, allows mixing of in phase signals between channels, i.e. L+R and R+L signals, the exact pro- Figure 1. Block diagram of one channel of the Consonant. Figure 2. Complete circuit of the left channel of the Consonant, including the power supply, which is common to both channels. The right channel is identical. V Figure 3. Frequency response of the Consonant with all filters and tone controls cancelled. Figure 4. Response of the scratch and rumble 7-40 — elektor july/august 1978 HIeseisS L|7o^B Sio- mam'&tEx EaSfll c'aflnii* || ■SMf hsIkIi IpiS B] ■«r EpTfll VSnjUl hHHS lSi\ g£/Bj sS«i P&£. 4 Jll imm saTER Ipa llj \mmm IiSmi^PF 1 ■: 1«" IHEI [;jg| n «¥! in E^OK ill elektor july/a portion of each signal that appears in the opposite channel depending on the setting of P3. With P3 set to minimum resistance complete mixing of both channels will occur and a mono output will result. When P3 is set to maximum resistance it will have little effect and the super-stereo image produced by antiphase signal mixing will result. Approximately mid-way between these two extreme positions the +R signal introduced into the left channel via P3 will cancel the -R signal introduced by R35, so only the L signal will appear in the left channel. Similarly only the R signal will appear in the R channel, i.e. a normal stereo image will be produced. Of course, it is also possible to obtain normal stereo image width by switching out the image width control using S4 so that there is no cross-connection be- tween the left- and right channels. Power supply The control amplifier requires a stabil- 7-42 - elektor july/august 1978 ised supply of approximately 21 V, which consists of zener diode D1 and T5 in figure 2. This also provides the supply to the disc preamp. Provision is also made for an IC regulator to provide the +15 V supply which will be necess- ary if a LED signal level indicator is incorporated into the circuit. This indi- cator may consist of the PPM and UAA 1 80 LED voltmeter described in Elektor 33, January 1978. Alterna- tively, the •Luminant’, a sophisticated LED meter which displays peak and average signal levels simultaneously, is described elsewhere in this issue. Performance The measured performance of the Consonant is illustrated in figures 3 to 8 and in the table of specifications. Fig- ure 3 shows the frequency response of the amplifier with all filters and tone controls cancelled. Figure 4 shows the same curve with the scratch and rumble filter characteristics superimposed. The response curves of the tone controls, showing the effect of the switchable turnover frequencies, are given in figure 5. The two graphs for crosstalk, left chan- nel crosstalk on right channel and vice versa, are given in figure 6. As is to be expected, channel separation is best at low frequencies and deteriorates towards the high-frequency end due to stray capacitance coupling between the two channels. Nevertheless, crosstalk at 1 kHz is a healthy -50 dB relative to 0 dB = 775 mV RMS at 1 kHz. Slight differences between the L on R and R on L crosstalk figures are due to inevi- table asymmetry in the left- and right channel layout on the p.c. board. Second-harmonic distortion is measured at three different levels in figure 7, — 10 dBm, 0 dBm and +10 dBm. Even at + 10 dBm, when the amplifier output is - 2.45 V RMS or 6.92 V peak-to-peak, the second-harmonic distortion does not exceed 0.23% at any frequency! At the normal operating levels of the amplifier -vsecond harmonic distortion is, as might be expected, considerably lower, less than 0.07% at 0 dBm and less than ! 0.04% at -10 dBm. furthermore, the distortion produced iy Consonant is predominantly the less ' >bjectionable second-harmonic, as fig- are 8 illustrates. Second-harmonic dis- tortion relative to 0 dBm = 775 mV is again shown in the upper trace, whilst the lower trace is the corresponding third-harmonic distortion, which is below 0.02% at all frequencies. Construction 4 C3, C2' and C3' should be tantalum types for low leakage, whilst IC1 and IC1’ must be type TL084. The temp- tation to use the cheaper, pin-compatible LM 324 should be resisted, as this 1C does not have FET inputs and has poorer performance. Power supply The Luminant operates from a sym- i metrical ± 15 V supply. The current | drawn from the negative supply between 15 and 25 mA, whilst that " I U Tirawn from the positive supply is about 25 mA plus 12 mA per LED, a total of about 170 mA with all LEDs lit. The ~ | circuit of a suitable power supply i given in figures 9a and 9b. Figure 9a shows how an unregulated positive and negative suppky may be obtained from a transformer having a single, untapped secondary winding, whilst figure 9b shows a simple stabiliser circuit. If the Luminant is used with the Consonant control amplifier then the 1 , arrangement of figure 9a may be used to obtain the negative supply from the i Consonant mains transformer. However, only the negative stabiliser need be used as the Consonant has provision for an on-board +15 V IC regulator. If the Luminant is used with a power amplifier having a symmetrical supply then the circuit of figure 9b may be connected direct to the supply rails of the amplifier. Finally, if the Luminant is used with a preamp or control amplifier having a ± 15 V supply, it may be connected direct to the preamp supply without the need for stabiliser circuits. Figure 7. Printed circuit board for the display drive circuitry (EPS 9949 - 2). Figure 8. Printed circuit board for the LED display (EPS 9949 - 3). Figures 9a and 9b. Power supply suitable for the Luminant. missing link Modifications to Additions to Improvements on Corrections in Circuits published in Elektor Elbug It is possible to make a small but signifi- cant improvement in Elbug, the monitor software routine for the Elektor SC/MP system. Elbug utilises location 0FE0 as an address flag for the cassette routine. However this location is also used by the modify routine, with the result that certain complications arise if one wishes to run the cassette routine at a speed of other than 600 baud. The end of a cassette transfer is no longer indicated by the word ‘Elbug’ appearing on the display, unless the data being transferred (from cassette to memory) is always accompanied by the start- and finish addresses. The above problem can be resolved by reserving a different location as address flag for the cassette routine. This can be done by modifying the contents of 02FF, 0353 and 0374 (these addresses correspond to locations 0FF, 153 and 174 in EPROM II). Until now, the contents of these addresses were 00, i.e. they were unprogrammed locations, which means that they can still be modified. By programming 10 into these locations, address 0FF0 will be reserved as address flag for the cassette routine. This address will remain unaf- fected by the modify routine. Once the above modification has been carried out, the cassette transfer routine will function satisfactorily at any desired speed. * * * RAM I/O When the RAM I/O card is used in con- junction with the CPU- and extension card, the value of resistors R14 . . . R21 should be reduced to between 470 and 1 k. Since the TTL ICs connected to the address bus cause excessive loading, the ‘0’ level can be as high as 1.4 V, This is remedied by altering the value of the above-mentioned resistors. coming soon piano oscillographics car start booster puffometer 24 dB VCF digiscope elektor july/august 1978 — 7 § 5=^ IC counter timebase A new Mostek IC, the MK 5009, is a complete timebase for frequency counters and other applications. The internal block diagram of the IC is shown in figure 1 . It incorporates a clock circuit (which may be used either with an external reference frequency, with an external RC circuit or with an external crystal) and a programmable divider. By applying an appropriate binary code to the programming inputs the division ratio may be varied in decade steps from 10° to 10®. Other division ratios are also available, the most interesting being divide-by- 2 x 10 4 , which gives a 50 Hz output with a 1 MHz clock frequency or a 60 Hz output with a 1 .2 MHz clock frequency. A practical circuit using the MK 5009 is shown in figure 2, whilst table 1 shows the division ratios that may be obtained for the different settings of SI to S4. A 1 MHz crystal is used in a parallel resonant circuit and this crystal should be a 30 p parallel resonant type. The divided down output is available at pin 1 of the IC. This output will drive CMOS circuits or one TTL unit load directly. However, since the IC is fairly expensive it is recommended that a permanent buffer stage be connected to the output. This may take the form of a TTL or CMOS gate or buffer, a FET or a bipolar transistor. The 1 MHz clock frequency is available at pin 1 0 of the IC and must similarly be buffered if it is to be used. Trimming of the oscillator frequency for maximum accuracy can be carried out using C2. With the division ratio set to 1 0, C2 can be adjusted for zero beat between the output and an accurate frequency such as the 200 kHz Droitwich transmitter. 7-56 — elektor july/august 1978 digital delay line There are several applications for a circuit that can delay digital signals, the Digital Reverberation Unit described in the recent May issue of Elektor being but one example. The most common method of delaying digital signals is to use a shift register. The circuit shown in figure 1 may therefore appear somewhat unusual, in that the two i most prominent components are not shift registers but Random Access Memories (RAMs; IC2 and IC3). In this circuit, the digital data are stored in the RAMs for the duration of the desired delay period and then recalled and presented at the ‘delay data’ output. Since the 1024-bit RAMs used here (type 2102) are relatively inexpensive, there is no need to skimp on memory space - if longer delay times are desired, the circuit can easily be extended as will 1 be described later. The circuit operates as follows. A clock signal is fed to three 4-bit binary counters, connected in cascade to produce a 1 2-bit counter. The first ten bits from this counter are applied to the address inputs of both RAMs. The eleventh bit is used to drive the read/write control input of the memories; inclusion of N3 in the feed to IC3 ensures that when IC2 is being read out (read mode) IC3 is storing the input data (write mode), and vice versa. In other words, the two RAMs are used alternately: when the contents of one are being scanned and fed to the output, the other is storing new data; when this cycle is completed the first RAM is used to store the incoming data and the second is read out. No logic gating is required to route the incoming data to the correct RAM: all data are presented to the data input of both RAMs. The memory which is in the Write mode will store the data, whereas the other will simply ignore them. The same is not true, however, for the data output: gates N1 and N2 are required to select the correct output at any given time. As mentioned earlier, the delay line can be extended by adding further RAMs. The alternating read/write operation used in this system involves using an additional pair of RAMs for each extension step. However, little more is required: the address inputs of the additional RAMs are simply connected in parallel with the existing address lines and the same read/write signal is * 0 applied. Gates N1 . . . N3 must be repeated for each further pair of RAMs. The ‘delayed data’ output of the first pair is connected to the ‘data input’ of the second pair, and so on down the line. This digital delay line can be used in the Digital Reverberation Unit mentioned earlier. Each pair of RAMs will replace one shift register IC. However, since the logic levels are not identical a simple interface circuit is required at both ends of the digital delay line. Figure 2 shows the principle: the output from IC3 in the original circuit is buffered by a single transistor and fed to the ‘data’ input of the delay line described here; the output from the delay line is again buffered by a single transistor and fed to the input of IC4 in the original circuit. elektor july/august 1978 — 7-57 audio mixer A novel approach is employed in this circuit, which allows mixing of two audio signals and cross-fading between them. Rather than using conventional potentiometers as analogue attenuators together with a summing amplifier, the circuit functions by sampling the two signals alternately at a high frequency. The input signals are fed to a pair of electronic switches each comprising two elements of a 4066 CMOS analogue switch IC. The use of a switch in shunt with the signal path as well as in series allows a high load impedance to be used (for low distortion) whilst at the same time maintaining good signal isolation when the switch is ‘off’. The two switches are opened and closed alternately by a 100 kHz, two- phase clock constructed around N 1 to N6. When SI is closed, S2 is open and signal A is fed through to IC 1 . S3 however, is open and S4 closed, so signal B is blocked. When S3 is closed and SI open then signal B is, of course, passed while A is blocked. PI allows the duty-cycle of the clock pulses to be adjusted, i.e. the proportion of the total time for which each signal is passed. This in turn varies the amplitude of each signal. With PI in its mid-position both signals have approximately the same amplitude whilst at the two extremes one signal is completely blocked whilst the other is passed continuously. A lowpass filter built around IC1 removes any clock frequency components from the output. Although these are inaudible they could, if not filtered out, damage power amplifiers and loudspeaker tweeters or beat with the bias oscillator in a tape recorder to cause high-pitched bleeping tones. The supply voltage, which should be stabilised and ripple-free, may lie between 9 V and 15 V. Above 15 V the CMOS ICs may be damaged and below 9 V the 741 will not function satisfactorily. The maximum input signal that the circuit will accept without distortion is about 1 V RMS. cheap crystal filter In view of the dramatic drop in the price of crystals used in colour TV sets, they now represent an economical way of building an SSB- filter. The circuit shown in the accompanying diagram is for a filter with a -6 dB bandwidth of roughly 2.2 kHz. The layout for the p.c.b. indicates 7-58 - elektor july /august 1978 Oo-ll IK3* -I — 1 — ± — 1 — — ® how such a circuit can be con- structed. This type of arrangement has the advantage that the input and output are as far as possible apart from one another, so that rejection outside the passband is at a maximum. By terminating the input and output with a 1 k resistor in parallel with an 1 8 p trimmer capacitor, passband ripple can be tuned down to 2 dB. The most important specs are given in the accompanying table, to which should be added that the out-of- band attenuation is 90 dB. Resistors: R1.R2- 1 k Capacitors: C1,C2,C4,C5 = 82 p C3 = 15 p C6,C7 = 100 n ceramic Miscellaneous: X1,X2,X3,X4. X5.X6 = 4.433,618 kHz Table. fo f_ 6dB < r > f-6dB ID *—60 dB ID f — 60 dB