1 LL 1-04 — elektor January 1977 Editor Deputy editor Technical editors Art editor Subscriptions W. van der Horst P. Holmes J. Barendrecht G.H.K. Dam E. Krempelsauer G.H. Nachbar Fr. Scheel K. S.M. Walraven C. Sinke Mrs. A. van Meyel UK editorial offices, administration and advertising: 6 Stour Street, Canterbury, CT1 2XZ. Tel. Canterbury (0227) - 54430. Telex: 965504. Bank: Midland Bank Ltd Canterbury A/C no. 11014587, Sorting code 40-16-11, giro no. 3154254 Assistant Manager and Advertising : R.G. Knapp Editorial : T. Emmens Elektor is published monthly on the third friday of each month, price 45 pence. Please note that number 27/28 (July/August) is a double issue, 'Summer Circuits', price 90 pence. Single copies (including back issues) are available by post trom our Canterbury office to UK addresses and to all countries by surface mail at £ 0.60 Single copies by air mail to all countries are £ 0.95 Subscriptions for 1977 (January to December inclusive): to UK ad- dresses and to all countries by surface mail: £ 6.25, to all countries by airmail £ 11.-. All prices include p & p. Change of address. Please allow at least six weeks for change of address. Include your old address, enclosing, if possible, an address label from a Letters should be addressed to the department concerned: TQ = Technical Queries, ADV - Advertisements. SUB = Subscriptions; ADM = Administration; ED = Editorial (articles submitted for publication etc.); EPS = Elektor printed circuit board service. For technical queries, please enclose a stamped, addressed envalope. The circuits published are for domestic use only. The submission of designs or articles to Elektor implies permission to the publishers to alter and translate the text and design, and to use the contents in other Elektor publications and activities. The publishers cannot guarantee to return any material submitted to them. All drawing, photographs, printed circuit boards and articles published in Elektor are copyright and may not be reproduced or imitated in whole or part without prior written permission of the publishers. Patent protection may exist in respect of circuits, devices, components etc. described in this magazine. The publishers do not accept responsibility for failing to identify such patent or other protection. National advertising rates for the English edition of Elektor and/or international advertising rates for advertising at the same time in the English, Dutch and German issues are available on request. Distribution: Spotlight Magazine Distributors Ltd, Spotlight House 1, Bentwell Road, Holloway, London N7 7AX. Copyright ©1977 Elektor publishers Ltd — Canterbury. Printed in the Netherlands. Deminer What is a TUN? What is 10 n? What is the EPS service? What is the TQ service? What is a missing link? Semiconductor types Very often, a large number of equivalent semiconductors exist with different type numbers. For this reason, ‘abbreviated’ type numbers are used in Elektor wherever possible: - '741 'stands for pA741, LM741 , MC741 , MIC741 , RM741. SN72741 , etc. - 'TUP' or 'TUN' (Transistor, Universal. PNP or NPN re- spectively) stands for any low frequency silicon transistor that meets the specifications listed in Table 1 . Some examples are listed below. - 'DUS' or 'DUG' (Diode. Uni- versal, Silicon or Germanium respectively) stands for any diode that meets the specifi- cations listed in Table 2. - BC107B', 'BC237B', 'BC547B' all refer to the same 'family' of almost identical better-quality silicon transis- tors. In general, any other member of the same family can be used instead. (See For further information, see TUP', TUN. DUG, DUS'. Elektor 20, p. 1234. Table 1. Minimum specifications for TUP (PNP) and TUN (NPN). '20V 100 mA 100 100 mW 100 MHz Some 'TUN's are: BC107, BC108 and BC109 families; 2N3856A. 2N3859, 2N3860, 2N3904, 2N3947, 2N41 24. Some 'TUr's are: BC177 and BC178 families; BC1 79 family with the possible exeption of BC1 59 and BC1 79; 2N2412, 2N3251 . 2N3906, 2N4126, 2N4291 . Table 2. Minimum specifications for DUS (silicon) and DUG (germanium). DUS DUG Vr, max iR.max Ptot, max 25V 100mA IpA 250mW 5pF 20V 35mA 100 pA 250mW lOpF Some 'DUS’s are: BA127, BA217. BA218, BA221. BA222, BA317, BA318, BAX13, BAY61, 1N914, 1N4148. Some ‘DUG's are: OA85, OA91 , OA95. AA116. BC107 (-8, -9) families: BC107I-8, -9I.BC147 (-8, -9), BC207 (-8, -9). BC237 (-8, -9), BC317 (-8. -9), BC347 (-8. -9). BC547 (-8, -9). BC171 (-2, -3). BC182 (-3, -4). BC382 (-3, -4), BC437 (-8. -9), BC414 BC177 (-8, -9) families: BC177 (-8, -9), 8C157 (-8,-9), BC204 (-5. -6). BC307 (-8, -9). BC320 (-1 , -2), BC350 (-1 . -2), BC557 (-8, -9), BC251 (-2, -3), BC212 (-3,-4), BC512 (-3, -4). BC261 (-2.-31.BC416. Resistor and capacitor values When giving component values, decimal points and large numbers of zeros are avoided wherever possible. The decimal point is usually replaced by one of the fol- lowing international abbrevi- p (pico) » 10" 13 n (nano-) - 10"* p (micro) =10"* m (milli-) = 10" 3 k (kilo-) = 10 3 M (mega-) = 10* G (giga-) = 10’ A few examples: Resistance value 2k7 : this is 2.7 kn. or 2700 Q. Resistance value 470: this is 470 n. Capacitance value 4p7: this is 4.7 pF, or 0.000 000 000 004 7 F . . . Capacitance value 10 n: this is the international way of writing 10,000 pF or .01 pF, since 1 n is lO"’ farads or 1000 pF. Mains voltages No mains (power line) voltages are listed in Elektor circuits. It is assumed that our readers know what voltage is standard in their part of the world! Readers in countries that use 60 Hz should note that Elektor circuits are designed for 50 Hz operation. This will not normally be a problem; however, in cases where the mains frequency is used for synchronisation some modifi- Technical services to readers — EPS service. Many Elektor articles include a lay-out for a printed circuit board. Some - but not all - of these boards are available ready-etched and predrilled. The 'EPS print service list' in the current issue always gives a complete list of available boards. — Technical queries. Members of the technical staff are available to answer tech- nical queries (relating to articles published in Elektor) by telephone on Mondays from 14.00 to 16.30. should be addressed to: Dept. TQ. Please enclose a stamped, self addressed envel- ope ; readers outside U.K. please enclose an I RC instead of stamps. — Missing link. Any important modifications to, additions to, improvements on or correc- tions in Elektor circuits are generally listed under the heading 'Missing Link’ at the earliest opportunity. contents elektor january 1977 - 1-05 Volume 3 Number 1 Despite its simple design the stereo audio mixer will accept inputs from microphone, magnetic cartridge, or 'flat' line sources. The noise generator described in this issue uses digital techniques to produce a pseudo-random binary noise signal. Suitable filtering converts this to analog 'white' or 'pink' noise. This photo of a Y-p reamp illustrates the modular construction of the Elektorscope. ulkhtih* ei Not a microphotograph of snow crystals, but neatly packed heat- sinks! selektor 110 elektoracle 113 stereo audio mixer 114 This is a design for a simple but high-quality, portable, five- input, stereo mixer intended for use in discos or for tape recording. marbles 120 Playing marbles is still one of the most universal of childhood pleasures. In this modern day and age. it was only a question of time before some designer tried his hand at designing an electronic equivalent . . . noise generator 123 A digital method of generating so-called pseudo-random 'white' and 'pink' noise is described in this article and applied in a practical circuit. audio wattmeter — J. Jacobs 128 elektorscope (2) 130 A linear timebase and stable trigger circuits are essential to the operation of an oscilloscope. In this month's article the circuit of the timebase, trigger circuit, X and Y amplifiers, channel switches and Y-preamps are discussed. quadi-complimentary complemented 139 bi-fetopamps 140 Interesting results can often be obtained by using several different types of active device in one circuit. A relatively new example of this is the BI-FET opamp. It consists of JFETs and bipolar transistors on the same chip, and this could well give interesting results at a reasonable price. ic audio 142 In this article two completely different types of high-power audio ICs are discussed. To avoid confusion the article is split into two parts. The first part deals with hybrid power- integrated circuits while the second section deals with the TDA2020 IC. the BF 494 — a case in point 148 transistors 149 market 150 linear ics, mos ics, ttl ics 157 tup-tun-dug-dus 160 advertisement elektor January ©fekwif hmk loon ms dir© s\ ovolable This is a selection from various issues: number 1 : @ 50 p - tup-tun-dug-dus - aqua amplifier - distortion meter - tap sensor - electronic loudspeaker - steam whistle number 2: @ 50 p - minidrum - universal display - dil led probe - big ben - modulation systems - how to gyrate number 3: @ 50 p - tap preamp - pll systems - fido - time machine - compressor - disc preamp - a/d converter - led displays number 4: @ 50 p - interference suppression - thief suppression in cars - supplies for cars - cybernetic beetle - quadro in practice number 5: @ 85 p 'Summer Circuits' issue, with over 100 circuits: amplifiers, generators, dividers, universal frequency reference, im- proved 7-segment display, receivers, power supplies, rhythm generators, measuring equipment, etc. number 6: @ 50 p - edwin amplifier versitale digital clock - phasing - disco lights number 9: @ 55 p — feedback pll for fm — function generator — racing car control — simple mw receiver — digital master oscillator — pll-ic stereo decoder number 10: @ 55 p — call sign generator — morse decoder — speech processor — morse typewriter — digital wrist watch the contents of the number 11: @ 55 p — tv tennis extensions — ssb receiver — tv sound front-end — dynamic noise limiter — integrated voltage regulators number 12: @ 55 p — ic rhythm generator — Polaroid timer — led meters — stylus balance number 13: @ 55 p — integrated indoor fm — equin (2) — digidisplay — versatile logic probe number 14: @ 55 p — a.m. mains intercom — fet probe — vhf fm reception — led light show number 15/16: @95 p ‘Summer circuits' issue, with over 100 circuits number 17: @ 55 p — m.p.g. indicator — ignition timing strobe — car service meter — windscreen wiper delay - rev counter number 18: @ 55 p number 19: @ 55 p - albar - metal detector - calendar - intercom - signal horn - pulse generator number 20: @ 55 p - thermometer - snooze-alarm-radio-clock - parking meter alarm - elektor index 1976 - phasing and vibrato - elektorscope Prices quoted @ U.K./ Surface rate. Airmail add. 35 o Following the success of the many original and W varied construction projects in Elektor J magazine we are now publishing X X^ BOOK 75. X Written and presented in X Elektor's unique X style its contents X include: X versatile digital clock mini hifi tv tennis game tunable aerial amplifier disc preamp universal frequency reference time signal simulator edwin amplifier, fido and many others PLUS 'DATA SECTION' covering Elektor ^ service to readers, LED DISPLAY V X MOS-ICS, TTL-ICS, OPAMPS, X X TRANSISTORS X X^ and X X TUP-TUN- X X DUS-DUG. X For ordering please use the postage paid order card in this issue. 1-14 — elektor january 1977 stereo audio mixar I This is a design for a simple but high-quality, portable, five-input, stereo mixer intended for use in discos or for tape recording. Despite its simple design the mixer will accept inputs from microphone, magnetic cartridge, or 'flat' line inputs. The total current consumption is quite low, so the unit may be powered from batteries as well as its own, integral, mains power supply. In the fields of application for which this mixer is intended it was felt that I I the cost, complexity and weight of such facilities as tone and balance controls were not justified. Such facilities are best left to the professional studio ! mixer. A block diagram of one channel of the mixer is shown in figure 1. This is of I course duplicated for the other channel. It can be seen that the first three inputs are each provided with a preamplifier. and any or all of these preamplifiers can be constructed as either a disc or micro- phone preamp. The outputs of the preamplifiers feed into faders (gain controls). The two remaining inputs are intended to be fed from ‘flat’ sources such as tape, and these inputs are fed direct into the faders. The outputs from all the faders are mixed (more properly ‘summed’) at the virtual earth input of the post ampifier, and the mixed output can then be fed Specification Five (stereo) inputs, three microphone or magnetic cartridge, plus two line inputs. Preamplifier Disc inputs x 80 (38 dB) Mic. inputs x225 (47 dB) Post amplifier ~ xl.25 Maximum input level (one input driven). Disc 44 mV RMS Sine- wave into 47 k. Mic. 14 mV RMS Sine- Line 2.5 V RMS into 47 k. Maximum output level 3.2 V RMS sinewave from 600 SI Frequency response Line and Mic.: 20 Hz - POkHz <— 2dB). Disc - RIAA equalisation correct to ± 1.5 dB. Total harmonic distortion Less than 0.1% nS'rSo Better than 70 dB. to the tape recorder, disco amplifier or I whatever. The post amplifier, according I to the whim of the individual construe- I tor, may or may not be equipped with a | master gain control. Preamplifiers Figure 2 is the complete circuit of one I channel of the stereo mixer. The input I preamplifier is constructed around T1 and T2, and is, of course, duplicated for I inputs 2 and 3, though this is not shown I in full. T1 functions as a voltage ampli- I fier with a relatively high gain, and the I noise figure is good due to the low 1 collector current of approximately I 86 iiA. T2 also operates as a voltage amplifier, I but since the output signal is taken from I its collector, the collector resistor must be fairly low to obtain a reasonably low output impedance that will not be unduly loaded by the fader PI . Negative feedback is provided from the . collector of T2 to the emitter of T1 via I an equalisation network connected be- I tween points X and Y. The networks for disc and microphone | inputs are shown at the bottom of I figure 2. Note that for the disc preamp I R3 is 470 S2, and for the microphone in- I put, lk5. The two line inputs require little com- I ment, since they simply feed direct into I the faders P4 and P5. It is, of course, I perfectly possible to convert any or all I of the preamp inputs to line inputs I simply by omitting the preamp com- I ponents and joining points A and B. Dl...D4=4x1N4001/ B40C400 The outputs of the faders are connected to the input of the post amplifier via mixing resistors R7 to Rll. To avoid any unwanted interaction between the input signals the a.c. voltage at the mixing mode must be zero. This is basi- cally what is meant by the term ‘virtual earth’: although the amplifier input is not actually grounded its input voltage is always very small. In other words the amplifier has a very low input im- pedance. This result is achieved by applying a large degree of negative feedback from the emitter of T4 to the base of T3. When one of the input signals swings positive this will attempt to force more base current into T3. The collector of T3, and hence the emitter of T4, will swing negative until the current through R13 is the same as that through the in- put resistor (neglecting T3 base current), and the a.c. voltage at the input node will remain zero. The overall gain of the post amplifier is the product of the gain of T3 (with feedback) and the gain of T4. This is i.e. about 1.25; Rj n is one of the resi- stors R7 . . . R1 1. The output of the post amplifier is taken from the collector of T4, the out- put impedance being 600 f 2. If desired R17 may be replaced by a master fader control, though in many cases this will not be necessary as the succeeding equipment will have gain controls. Power Supply The power supply is based on a 723 IC regulator, with its output set to 12 V, and this is more than adequate to supply the small current taken by the mixer without the need for heatsinks or external transistors. Due to tolerances in the 723’s internal reference voltage the output voltage may not be precisely elektor january 1977 — 1-19 5 ■hi ■in IIIIH ■III ■III HIM HIM ■III HIM HIM nv ■III IIIIH inn ■ III ■III HIM IIIIB ■III Him ■hi HIM ■ IIIMB ■III M ■hi HIM ■III HIM ■III HIM ■II Photo 3. The completed mixer in its case with the lid removed. figure 3. Power supply for the stereo mixer. figure 6. Frequency response of the mixer for the disc (fed via reciprocal RIAA network) microphone and line inputs. 1 2 V, but this is no cause for concern. Construction A printed circuit board and component layout for the mixer are given in figures 4 and 5. It will be seen that the input preamplifiers are duplicated in three pairs, each numbered identically, the right channel components being identified by an apostrophe. When ordering components it must be remem- bered that for the input preamps 6 off are required of each component, and for the post amplifier, 2 off. For the faders, slider pots of the RS Components, Doram or Radiohm type should be used, as these may be mounted direct on the back of the p.c. board as shown in photos 1 and 2. It should be noted that the sliders for the line inputs are mounted on the back of the board adjacent to the power supply, on either side of IC1. Provision is made for mounting the mains transformer on the p.c. board, and this should prove satisfactory. However, if hum on the inputs is a problem it may be necessary to mount the transformer remote from the board. Input and output wiring should conform to normal audio practice, care being taken to avoid earth loops. Performance and Applications Figures 6a, b, and c show respectively the frequency response of the disc, microphone and line inputs. It would have been extremely easy to have extended the h.f. response into the MHz region, but bearing in mind the probable applications it was felt that this was simply inviting problems due to radio breakthrough and instability, especially in such applications as discos and tape recording where the layout of leads may be somewhat haphazard. For this reason the response is confined strictly to the audio spectrum by the inclusion of C6, and Cl 2 in the micro- phone preamp. The mixer may be used with most types of magnetic cartridge and dynamic microphone, as well as with high-level sources such as cassette decks, tape decks and tuners. With high output cartridges it may be found that over- loading of the preamp occurs, and the cure for this is simply attenuation of the cartridge output so that the preamp reaches maximum output only on peaks. M 1-20 — elaktor january 1977 marbles Playing marbles is still one of the most universal of childhood pleasures. In this modern day and age, it was only a question of time before some designer tried his hand at designing an electronic equivalent . . . There are any number of different games that can be played with marbles. This electronic version simulates one particular variant: the object is to roll the marble towards a wall, and the person whose marble ends up closest to the wall is the winner. It doesn’t matter whether it bounces off the wall first - the only thing that counts is the final position. In the electronic version, the rolling marble is simulated by an up/down counter driving a column of LEDs (figure 1). It is ‘rolled’ by pressing the start button; the length of time the button is depressed corresponds to the initial speed and this is where some skill is called for. Initially, LED 1 is on. As soon as the button is pushed this LED goes out and LED 2 lights up; then this, in turn, goes out and LED 3 lights and so on. The speed with which the point of light moves up the column depends on how long the start button is held down. Since real marbles don’t keep rolling at the same speed all the way, some ‘friction’ has been built in: the point of light slows down gradually until it finally comes to a standstill. One LED is now on, corresponding to the final position of the marble. The three LEDs across the top of the column are the ‘wall’ against which the ‘marble’ bounces, so the object of the exercise is to end up with LED 16 lit up. Pushing the reset button resets the counter, so that LED 1 is lit up ready for the next try. The circuit The basic operation of the circuit is as follows. When the start button is depressed an oscillator is started. The, initial frequency depends on how long the button is held down, and after it is ; released the frequency of the oscillator | gradually decreases until it finally stops. The output from the oscillator drives the up/down counter (see figure 1 ). The four-bit output from this counter is passed to a decoder which drives the LEDs. The complete circuit is shown in fig- ure 2a and 2b. T1 and T2 are the oscil- lator — in this case, a fairly standard multivibrator circuit. The frequency is -22 - alektor ja 1977 marbles ‘marble speed’ control. The output at the collector of T2 is not a particularly good square-wave, so it is ‘polished up’ by two Schmitt-triggers (N22 and N21) in cascade. The up/down counter (1C8) has two inputs, one for ‘count up’ and one for ‘count down’. Which of these inputs is driven by the oscillator depends on the state of the RS-flip/flop N3/N4. When the output of N3 is at logic ‘1’, N7 is freed and the counter counts up; when the output of N4 is ‘high’, the counter counts down. When the reset button is depressed, the flip/flop is reset to the condition where the output of N3 is T’. The counter will now count up until it reaches the maximum count (1111). At this point the output of N23 changes from ‘1’ to ‘O’. This low-level input to N4 sets the flip/flop. The output of N4 is now ‘high’ and the counter counts down. When it reaches the lowest count (0000) the output of N24 goes ‘low’, resetting the flip/flop again so that the counter starts to count up again. The result of all this is that the counter counts up and down continuously until the oscillator stops. The ‘4 - to - 16 decoder’ and the LED drivers are shown in figure 2b. The number of components required has been reduced by incorporating two tricks: the LED drivers are part of the decoder, and the decoder actually consists of two ‘2 - to - 4’ decoders instead of one ‘4 -to - 16’ type. For a particular LED to light up, the left-hand transistor connected to its anode and the right-hand transistor connected to its cathode must both be conducting. The left-hand transistors are driven by the A and B outputs of the counter, as decoded by N13 through N16, and the right-hand transistors are driven by the C and D outputs through a similar decoder (N17 through N20). The transistors are connected to the LEDs in such a way that the final result is the desired ‘4 -to -16’ decode. To give one example, if the counter output is ‘1100’ ( A = 0, B = 0, C = 1 and D = 1 ) corresponding to a count of 12, the thirteenth LED should light up - remember that ‘0000’ corresponds to LED 1! In this case the output of N13 will be low, turning on T4. At the same time the output of N20 will also be low, turning on T1 1. T4 is connected to Dl, D5, D9 and D13, whereas Til drives Dl 3, D14, D15 and D16.The only LED connected to both transistors is D 13, so it lights up, as intended. LEDs D17, D18 and D19 are the ‘wall’. N lise generator elektor january 1977 — 1-23 Usually, noise is something we want to get rid of. However, there are applications in which noise is turned to practical use, such as for measuring and testing audio systems. A digital method of generating so-called pseudo- random 'white' and 'pink' noise is described in this article and applied in a practical circuit. Noise is one of the oldest known and still the most fascinating signals in mod- ern electronics. It is a signal which ap- parently varies at random with, time. ‘Apparently’, because certain laws of probability and statistics are obeyed. These mathematical backgrounds pro- vide us with quantities which can accu- rately define such a signal. They say I something about the probability of a I certain value being assumed, and become relevant only after the noise ; signal has been observed during a long i period. An example of such a mathematical i quantity is the mean square value of the noise voltage, v 2 . The root is the root mean square (RMS) value of the noise voltage. White noise For one class of noise signals the mean square value of the noise voltage is defined by: v 2 = c A f where Af = f 2 - f, , the difference between the highest and the lowest frequency of the frequency band under consideration (see figure 1). The factor c says something, or rather every- thing, about the power density spectrum. If c is frequency-independent, the power density spectrum is constant. This means that all frequencies are then represented, equally in the noise. The quantity v 2 increases with the band- Figure 1. The frequency spectrum of a band- limited 'white' noise signal. The r.m.s. value of the noise voltage in a range Af between f, and f, is proportional to the square root of Af. Figure 2. An n-bit shift register in combi- nation with suitable EXOR-feedback will produce pseudo random binary noise. width Af. This type of noise signal, where c is constant, is called ‘white noise’. The term ‘coloured noise’ is used for signals where c is frequency-dependent within the frequency range under consideration. Note that in practice, due to physical limitations, noise can only be ‘white’ in a band-limited frequency range. If Af were to become infinitely large while c remained constant, v 5 and the signal power would also become infinitely large. Pink noise One form of coloured noise is ‘pink’ noise. Both pink and white noise are useful in audio and acoustic measure- ments. To determine the frequency response of a system, a suitable noise signal is fed to the input. When the out- put is passed through a band-pass filter with a given central frequency and bandwidth, the r.m.s. value of the noise voltage corresponding to that particular frequency band will be measured. By using several band-pass filters with suit- able quality factors and central frequen- cies, the entire relevant frequency response of the system can be measured. 1. For filters with a constant bandwidth the quality factor Q increases pro- portionally with the central frequency f c . If a noise signal is measured with bandfilters of this type, the r.m.s. value of the output voltage measured for white noise is constant for each bandfilter. 2. For filters with a constant quality factor Q the bandwidth B increases with the central frequency f c . A well- known example of this type are the so-called ‘octave filters’, where the ratio between the highest and the lowest bandpass frequency is 2. ‘Third octave filters’ are also in common use. If white noise is applied to a system with a ‘flat’ frequency response, selec- tive measurement at the output with this type of filter will show that the r.m.s. value of the measured voltage increases in proportion to the square root of the central frequency. This is equivalent to saying the frequency characteristic rises at + 3 dB per octave. To correct for this, it is necessary to include a low-pass filter with a slope of - 3dB per octave. This filter is connected between the white noise generator and the 2 ZLT ZL IzJ FFj — FF3 °3 kJ H FFn-l H k 0 lo=Qn»Qm+Qn»Qr — , 1 Table 1 Tk 0 3 6 7 8 9 10 13 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 50 51 52 53 54 55 56 57 58 59 60 61 62 63 Q0 Q1 Q2 Q3 Q4 Q5 Q6 0 111 0 0 11 0 0 0 1 0 0 0 0 0 0 0 0 10 0 0 0 10 0 0 0 10 0 0 0 1 0 0 0 0 10 0 0 110 0 0 110 0 0 11 0 0 0 1 10 0 0 0 10 0 10 10 0 10 1 0 0 10 10 0 1 110 0 1110 1111 0 111 10 11 0 10 1 0 0 10 0 0 0 1 10 0 0 110 0 1110 0 111 0 0 11 10 0 1 0 10 0 0 0 10 10 0 1 0 10 0 10 10 110 1 0 110 10 11 110 1 1110 0 111 10 11 110 1 0 110 0 0 11 10 0 1 110 0 0 110 10 11 0 10 1 10 10 0 10 1 10 10 110 1 1110 1111 1111 1111 0 111 111 63 111 62 111 60 111 56 011 48 0 0 1 32 0 0 0 1 0 0 0 2 0 0 0 4 10 0 8 0 10 16 0 0 1 33 0 0 0 3 0 0 0 6 10 0 12 110 24 011 49 0 0 1 34 0 0 0 5 10 0 10 0 10 20 10 1 41 0 10 19 0 0 1 39 10 0 15 110 30 111 61 111 58 011 52 10 1 40 0 10 17 0 0 1 35 0 0 0 7 10 0 14 110 28 111 57 011 50 0 0 1 36 10 0 9 0 10 18 0 0 1 37 10 0 11 0 10 22 10 1 45 110 27 011 55 10 1 46 110 29 111 59 011 54 10 1 44 110 25 011 51 0 0 1 38 10 0 13 110 26 011 53 10 1 42 0 10 21 10 1 43 0 10 23 10 1 47 110 31 111 63 Table 1. The truth table for one complete cycle of a 6-bit shift register with EXOR-feed- back from the outputs 05 and Q6. Figure 3. The frequency spectrum of pseudo random binary noise. Figure 4. The pulse diagram corresponding to table 1. Figure 5. The circuit diagram of the white noise generator. N9 . . . N29. N3 and N4 are needed to terminate the 'all-zero' output condition. system. The noise signal at the out- put of this filter is called pink noise. How do we make noise? This may sound like a ridiculous ques- tion considering the fact that the major problem in audio is usually how to get rid of it. However, what is meant is: how can we make noise with the greatest possible spectral purity? And that is something quite different. White noise can be derived from the noise produced by the BE junction of a transistor that is reverse-biassed into the break-down region. This produces a very low-level signal which must be ampli- fied, and the y - noise of the amplifier may be a problem. The signal-to-noise ratio of the amplifier must be very high. An even greater problem is that this type of noise is of thermal origin. The noise (r.m.s. value) is sensitive to tem- perature variations. Pseudo random noise Pseudo random noise, unlike random noise, consists of a periodical repetition of a specific random noise pattern. As an example, imagine that a true random noise signal is recorded on a closed loop of tape. When this tape is played back. the output will be pseudo random noise, An advantage of using pseudo random noise as a test signal is that the meas- uring time need not be infinitely long. (Theoretically, this is a requirement when measuring with true random noise . . . ). Pseudo random binary noise It is not too difficult to generate pseudo random noise using digital techniques. A pseudo random binary noise signal is passed through a suitable low pass filter to give the analog noise required. It will be shown that both the binary and the analog noise signal are sufficiently ‘noisy’. The cycle character of pseudo random noise results in a repetition rate To of the analog noise (To is the period of one noise pattern), and as a cycle time T 0 for the binary noise. Figure 2 gives the block diagram of a binary pseudo random noise generator. The n flipflops FF! ... FF n are cas- caded thus forming an n-bit shift regi- ster. The register is shifted by a clock pulse with period time Tfc and frequency ffc. Since the continuous clocking of only ‘zeros’ or ‘ones’ into the shift register will certainly not lead to the required I noise generator binary noise signal, the first flipflop 1 FF j should receive a digital signal which is in some way related to the logic state of the register. The input of FFi re- ceives a signal Qo which is related to the signals Q n and Q m , according to the fol- lowing truth table: Qn Qm Qo 1 0 F 0 1 l 0 0 o 1 1 0 t The objective now is to choose values for n and m such that the maximum number of different output states of the flipflops FF, ... FF n is achieved. This corresponds to the maximum cycle time of the shift register. This requires some further explanation. The shift register consists of n flipflops, and each of the n Q-outputs can be considered as one bit of an n-bit binary number represented by the contents of the shift register. If the n flipflops are read out in parallel in the sequence Q n , Q n -i . . . Qi , we get an n-bit number corresponding to a given decimal number. The state of Qi corresponds to the value of the least significant bit (LSB) *), the state of Q n corresponds to the most significant bit (MSB) *). ') The concepts LSB and MSB indicate the place of the bit in the number. They are a measure of the weighting factor allocated to the value of the bit concerned. The MSB is to the extreme left and in the case of an n-bit binary number it indicates (0 or 1) x 2 n '' . The LSB indicating the units (0 = even. 1 = odd), is to the extreme right, and has the lowest weighting factor for whole numbers. In the decimal number 1976, digit 1 is the MSB (thousands), and 6 the LSB (units). The whole can be regarded as a counter that counts in a seemingly random sequence. The choice of m should be such, that all possible binary ‘numbers’ occur within a corresponding number of clock pulses. All possible numbers occur only once within the cycle period To. This maximum number of logic states is N = 2 n -l. This is one less than the absolute maximum because an ‘all-zero’ output has to be avoided, since this would otherwise continue infinitely. The (maximum) cycle time To, corre- sponding to this maximum number of states N is: T 0 = NT k . If, for a given number of flipflops n, feedback takes place from Q n and a randomly selected Q m , there is a good chance that the cycle will be shorter than the maximum cycle time NT k . It is very difficult to find values of m, corre- sponding to the selected n, in such a manner that the cycle time is maximum. Fortunately this work has already been done for us. Table 2 shows which outputs of the register should be used for the EXOR- feedback, for registers up to a maximum length n = 33. The last column also gives the corresponding cycle time expressed in clock periods T k . From table 2 it can be seen that there are always at least two possibilities: either Qn-m or Qm can be used. In table 2 Qn-m is shown in brackets. In a number of cases it is necessary to have EXOR-feedback from four Q-out- puts. For n = 8, for example, the feed- back condition is: Qo = Q 2 ©Q 3 ©Q4©Q« The © sign indicates the exclusive OR- function. Table 1 shows the truth table for a 6-bit shift register with a cycle time of 63 clock periods. The number in the first column gives the order in which the ‘random’ numbers appear at the output. The second column lists the EXOR in- formation Q 0 which is fed to the first flipflop. The next six columns indicate the output from the 6 flipflops and the last column gives the decimal value of the 6-bit binary number, where Qi represents the LSB and Q 6 the MSB. Table 1 and the pulse diagram of fig- ure 4 both clearly illustrate that the cir- cuit is basically a shift register: the bits move one step to the right at each clock pulse. After 63 clock pulses the register is back at its initial state. The pseudo random character of the output states is expressed as follows: 1 . Of the total of N clock periods in one cycle, any particular Q output is ‘high’ during (N+l) :2 periods, and low during (N-l) :2 clock periods. This is because the zero state of the register is excluded. As n, and with it N, increases, the chance of a logic ‘0’ approaches the chance of a logic ‘1* (50%). 2. If by traject we mean the number of clock periods within which the logic state of a particular Q output does not change, there are (N+l) : 2 tra- jects in each complete cycle. Half of these trajects are equal to one clock period; one quarter are equal to two clock periods, one eighth are equal to three clock periods, and so on. There is also one traject that is equal to n clock periods. 3. The number of trajects of each length is equally divided over trajects with logic ‘0’ and ditto with logic ‘1’; the Resistors R1 =390n R2, R3, R4. R5 = 1 k R6= 100 k capacitors Cl = 4n7 C2 = 6n8 Cx-2m2/16V semiconductors IC1 = 74132 IC2, IC3, IC4, IC5 - 7496 IC6. IC7, IC8, IC9 = 7405 IC10 = 7400 Table 2. A survey of possibilities and connec- tions for feedback shift registers with bit lengths ranging from 3 to 33. Figure 6. The printed circuit board and component layout for the circuit of figure 5. Figure 7. The circuit of a passive low-pass filter with a slope of 3 dB per octave. If white noise is applied to this filter, pink noise is pro- duced at the output. Figure 8. Frequency response of the filter of traject of n-l clock periods occurs in I ‘0’ only and the traject of n clock! periods occurs in ‘1 ’ only. These rules I can be verified by means of table 1 | and figure 4. The design described here (figure 5) uses a 20-bit shift register with EXOR feed- back from the outputs Qn and Qjo- The cycle duration is 1,048,575 (see table 2). It would be possible to set up the truth table, as was done for the 6-bit I register in table 1. However, such a truth table would comprise 23 columns I instead of the 9 in table 1 , meaning that I it would be nearly three times as wide. The number of lines becomes 1 ,048,576 i instead of the 64 of table 1 , which [ means that the truth table becomes more than 16,000 times as long. Since table 1 occupies a full column of an Elektor page, this truth table would cover over 16,000 pages. So as not to make this article unnecessarily long, we have refrained from publishing the table < in question. . . . The cycle time T 0 of the above-de- scribed shift register counter with maximum cycle time can be extended ] considerably. At n = 33 (see table 2) j and a clock pulse frequency of 1 0 MHz ] 0 k =0.1 ms) it takes about 859 seconds I (almost 15 minutes) before the cycle is I completed. If, instead, a clock period of j 1 second is used (ffc = 1Hz), the cycle I time would be equal to about 8.6 x 10 9 I seconds, considering that one year takes | on average 60 x 60 x 24 x 365.25 = 31,558 x 10 6 seconds, it would take | generator elektor january 1977 — 1-27 over 270 years to complete the cycle! Chances are that this exceeds the life of the hardware constituting the equip- The frequency spectrum Figure 3 is an attempt to illustrate the power density spectrum of the pseudo random binary noise on the Q-outputs. This spectrum is not continuous, but | consists of an infinite number of lines. The spacing between the lines (fre- quency difference) equals so that the spectrum consists of the fre- quencies The envelope of the frequency spectrum varies according to the function x being related to the ratio between f and ffc. The spectrum contains no com- ponent for ffc and multiples thereof. It can be calculated that the power spectrum has dropped 3 dB at a fre- quency equal to 0.45 x the clock fre- quency ffc. The spectrum is equivalent to a band- limited white noise signal (within 0.1 dB) for frequencies between 0 and fk/47r Hz. To be on the safe side, the band can be limited to For the final design fg was set at about 25 kHz, so that a clock frequency ffc of about 500 kHz is needed. Since N equals 1,048,575, the distance between the lines of the frequency spectrum is slightly less than 0.5 Hz. The cycle time To is about 2 seconds. The circuit ! The final circuit is shown in figure 5 . Schmitt trigger gates N1 and N2 are used as a clock pulse generator which runs at 500 kHz. This drives the clock inputs (pin 1) of four 5-bit shift regis- ters IC2 . . . IC5 ; these are cascaded to form the 20-bit shift register. The EXOR function is achieved by means of the gates N5 . . . N8. Feedback is taken from Q17 and Q20, in accord- ance with table 2. The EXOR feedback signal (the output of N5, Qo) is fed to the input of the shift register via in- I verter N3 and gate N4. Point 1 of N4 is normally ‘high’ so that the feedback signal arrives at point 9 of IC2. However, if all 20 Q-outputs of IC2 . . . IC5 are ‘low’, the input of in- verter N29 goes high. As a result point 1 of N4 goes low. A ‘1’ is now fed into the register so that the condition of twenty zeros, which might occur due to faults, is terminated. The output is fed through a low-pass filter (R4, R5, C2) with a cut-off fre- quency fg of about 25 kHz and a slope of 6 dB per octave. The output im- pedance is about lk; the peak value of the white noise voltage at the output is about 4 V. It can be calculated that this peak value is about 3.16 times the r.m.s. value. Consequently, this noise signal is emi- nently suitable for testing loudspeakers and amplifiers. If the drive level is set so that no ‘clipping’ occurs (this is audible as a change in the timbre of the noise), the amplifier will deliver only 10% of its maximum output power. This will usually mean that the loudspeaker is in no danger of falling victim to the measurements. To conclude with, figure 7 gives a cir- cuit for a 3 dB per octave low-pass filter. This can be connected to the out- put to convert the ‘white’ noise into ‘pink’. It is strongly recommended to use 5% capacitors and resistors. Figure 8 shows the frequency response of this filter. M 1-28 — elektor january 1977 audio wattmeter JJacobs Tantalized by the scent of more watts, potential buyers are often seduced by the over-glowing specifications that some manufacturers give for their amplifiers. On closer investigation, for instance, a '2 x 30 watt' amplifier may only by quoting music power, with the real power at no more than 20 watts sine power per channel, and with both channels being driven simultaneously the maximum power turns out to be only 2x14 watts. By means of the wattmeter described in this article, it is possible to determine the real output power of an amplifier (up to 140 watts per channel into a resistive load). 2 0 - D> Tt, 3 0 - > the rectifier and the meter often leads to inaccurate results. Figure 2 shows a method where the output of the amplifier is connected directly to the load resistor R. This resistor is shunted by a rectifier and moving-coil instrument in series. The power can be read directly from the meter. The same comments that apply to figure 1 are also valid for this system, with the extra disadvantage that the indication is non-linear, so that a custom-calibrated scale is required. Figure 3 shows a third possibility where an electronic AC voltmeter is connected in parallel with the load resistor R. This allows a high degree of accuracy because the non-linearity of the rectifier and the meter can now be compensated for by the electronics. The drawback with this system is that the output power is not presented as a direct reading: it It is possible, of course, to calibrate one or two scales corresponding to specific resistance values (4 and 8 J2, for in- stance). From the above, it can be concluded The important figure in an amplifier’s power specifications is the total output power with both channels driven. One channel on its own may deliver more power than with its partner if the power supply is inadequate. Also remember that it is unwise to test an amplifier beyond its specification, in case the heat sinks, for instance, are only suf- ficient for running within the specified limit. The actual output power of an amplifier can only be determined accurately by using an audio tone generator (signal source), an oscilloscope (to inspect the waveform), and an audio wattmeter. The wattmeter described here has been designed to have a wide frequency range (5 Hz-400 kHz) and is suitable for use up to 140 watts (into either 4 or 8 ohms). Various Measuring Methods The output power can be measured in various ways, as shown in figures 1,2 and 3. In the arrangement of figure 1, the output signal of the amplifier is rectified by a bridge rectifier and then fed to a load resistor (R) via an ammeter. A voltmeter is connected in parallel with the load resistor. The output power is calculated by multiplying together the current and voltage values given by the meters. The advantages of this method are its simplicity and wide frequency range; furthermore, it is only necessary to measure DC voltages and currents. However, the non-linearity of Figures 1 , 2 and 3. Various ways of accurately determining the output power of an amplifier. Figure 4. Detailed circuit diagram of the wattmeter, two of these circuits being re- quired for stereo equipment. Figure 5. A simple symmetrical power supply suitable for use as the current source for the wattmeter. audio wattmeter elektor january 1977 — 1-29 that each method has its drawbacks. However, the method shown in figure 3 has fewer disadvantages that the others, so it was chosen for the wattmeter described here. The circuit The amplifier under test is loaded by several resistors connected in series/ parallel (figure 4). Instead of using high power 4 and 8 £2 resistors, which would be difficult to obtain, these values are approximated by the resistor networks , shown. If 1 7 W types are used, the total power dissipated can be anything up to 140 W. Don’t mount these resistors too close to each other or to a base board: they can get very hot! It is | recommended that non-inductive re- sistors are used here - normal carbon resistors will do quite well. The load resistance is selected with S2, which should preferably be a make- before-break type. Note that the pos- ition marked ‘600 £2’ only gives 600 £2 load impedance when S3 is in position 1 . This facility has been included for dB measurements, where 0 dB corresponds to 0.775 V across 600 £2 (1 mW). S3 is the range selector switch. With the resistance values shown, the ranges will be sufficiently accurate. The ‘9 k’ resistor is actually two 18 k resistors connected in parallel. The opamp is connected in a standard AC voltmeter configuration. The trick is that the opamp tries to maintain the same volt- age on both its inverting and non- inverting inputs. The non-inverting input is connected to the input, and the inverting input ‘sees’ the voltage across the series connection of Rl, PI and Cl . The opamp will therefore produce an output current that builds up exactly the right voltage (i.e. the input voltage) across this series connection. The meter is connected in a bridge rec- tifier circuit in series with the opamp output. The output current must flow through the meter, and any non- linearities of the diodes (or meter) are compensated for by the opamp. The (rectified AC) current through the meter is equal to provided the impedance Cl is negligable when compared to Rl + PI, at the lowest frequencies involved. This ca- pacitor should be a non-polarised type or, if this is unavailable, two 1000 n capacitors connected ‘back-to-back’ as shown in the diagram. The 47 n capaci- tor is necessary to maintain a low impedance at high frequencies. The pin numbering shown for the opamp refers to the TO or mini-DIL case style. The bandwidth is set by the compensation networks R2/C2 and C3. With the values shown it will be approximately 400 kHz. Power Supply The supply voltage used here for the opamp is approximately ±10 V, The current consumption of the circuit (even in the stereo version) is so low that a very simple supply is sufficient. Figure 5 gives a simple, symmetrical supply which is suitable. The two 1 00 n capacitors are for h.f. decoupling, and although they are shown in the power supply here, they should be located as close as possible to the opamp. Calibration The wattmeter is calibrated with PI. The procedure is as follows. 1 . Set S2 in the ‘600 £2’ position. 2. Set S3 in position 2. 3. Temporarily short out Cl. 4. Connect a 2k2 resistor between the ‘+10 V’ supply and the input. 5. Adjust PI for full scale deflection (e.g. 3 Watts on the 4 £2 scale). 6. Remove the resistor and the short across Cl. It should be noted that the scale is not linear - it is square-law. As an example, if there are 10 divisions on the original (linear) scale on the meter, and full scale is now ‘300’, the original markings will now correspond to ‘O’, ‘3’, ‘12’, ‘27’, ‘48’, ‘75’, ‘108’, ‘147’, ‘192’, ‘243’ and ‘300’. Furthermore, when using the instru- ment one must realise that the scale calibration is only valid for a sine-wave input. If the sine-wave sounds audibly distorted, the actual power output can be either higher or lower than the value indicated. In general, one should not consider this unit to be ‘precision instrument’. This is one of the reasons why precision resistors are not specified. However, no- one should worry about the difference between, say, 100 W and ‘only’ 95 W - it just is not audible. 1-30 — elektor january 1977 elektorscope (21 A linear timebase and stable trigger circuits are essential to the operation of an oscilloscope. Good linearity of the timebase ensures that there is no distortion of the displayed waveform along the X-axis, while stable triggering ensures a jitter-free display. In this month's article the circuits of the time- base, trigger circuit, X and Y amplifiers, channel switches and Y-preamps are discussed. For the benefit of the less experienced reader the general principles of timebase and trigger circuits will briefly be dis- cussed. The timebase output consists of a linear ramp or sawtooth voltage which is applied to the X plates of the oscillo- scope, causing the trace to sweep hori- zontally across the screen in a linear fashion. Once the trace has been de- flected across the entire screen width the ramp voltage drops rapidly to zero and the trace returns quickly to the •starting position. To prevent the flyback or retrace from appearing on the screen the CRT beam current is cut off during this period, as mentioned earlier. During the sweep the trace is deflected in a vertical direction by the signal applied to the Y plates, thus causing the input waveform to appear on the screen. If the timebase were allowed to free run with no trigger input then it is likely) that the sweep would not start every time at the same point on the signal waveform. The portion of the signal waveform displayed during each sweep would thus be different, and the trace would appear to run one way or the other across the screen (figure 1 A). In order to obtain a stable trace the timebase must not be allowed to free run but must be started at the same point on the waveform for each sweep. This is shown in figure IB. The trigger circuit detects the amplitude of the waveform and also whether the slope is positive or negative-going - it would not do to have one sweep of the time- base triggered on a positive slope and the next sweep (at the same level) on a negative slope, since this would give rise to a mixed trace on the screen. Where successive cycles of a waveform are of the same amplitude the trigger level is relatively unimportant and it is common practice to trigger near the zero-crossing point so that the trigger point does not vary if the signal ampli- tude changes (figure 1C). However, if the amplitude of successive cycles was not the same then triggering at the zero- crossing point would mean that success- ive cycles of differing amplitude would appear at the same point on the screen. » In this case the timebase must be triggered only on the highest amplitude cycle of the waveform (figure ID). The trigger circuit is thus provided with a trigger level control to ensure reliable triggering on any repetitive waveform. Figure 1. Showing the principle of a triggered timebase. Figure 2. Block diagram of the Elektorscope timebase and trigger circuit. Figure 3. Diagram of the trigger circuit. Figure 4a. Circuit of the timebase. 1-32 — elektor january 1977 elektorscope (2) Figure 2 shows a block diagram of the trigger circuit and timebase. The output of the Y1 or Y2 preamp or an external signal may be selected as the trigger source. The trigger signal is compared with a continuously variable reference voltage (trigger level control). When the signal level exceeds the trigger level the output of the comparator goes high, and when the signal level falls below the trigger level the comparator output goes low. A polarity selector determines whether the timebase shall trigger on the positive or negative-going edge of the compara- tor output. The selected edge triggers a monostable that produces a short, fixed- length pulse which triggers the sweep generator. Finally, the output of the sweep generator is buffered by an out- put stage which drives the X amplifier. With the timebase in the ‘automatic’ mode the timebase will free run in the absence of a trigger signal. This is par- ticularly useful when observing DC levels which provide no trigger signal. Trigger circuit The complete trigger circuit is given in figure 3. T1 provides a fairly high input impedance and a gain of 4.7 to the input signal. The output from the collector of T1 is fed into the inverting input of an LM3 1 1 high speed compara- tor, while the non-inverting input is fed from the slider of PI to provide the trigger reference level. A small amount [ of positive feedback is applied around the comparator via R12 to provide a ! regenerative or ‘Schmitt Trigger’ type of I action to avoid trigger jitter on noisy I waveforms. Triggering of the timebase is performed by the upper of the two monostables i IC3. This can only be triggered by a positive-going pulse, so this makes for easy selection of trigger polarity. With S5 in the ‘pos.’ position the comparator output is routed through N4 and N3 so ! that the timebase is triggered on a positive-going edge. With S5 in the ‘neg.’ position the output of the comparator is inverted by N1 (giving a positive pulse on the negative-going edge of the com- parator output). The output of N1 is routed through N2 and N3 to the B input of the monostable. When the monostable is triggered it provides a short negative-going pulse to trigger the timebase. The second monostable is associated with the auto free-run facility. When a trigger signal is present this monostable > is continuously retriggered and its Q output remains high. In the absence of a trigger signal the monostable will reset and the Q output will go low, thus continuously triggering the timebase when S4 is in the auto position. The trigger circuit may be inhibited and the timebase switched into a continuous free-run position, thus grounding the trigger input of the timebase. Timebase The timebase (figure 4a) makes use of the well-known 555 timer. Those un- familiar with this IC should read the following description in conjunction with the internal diagram of the 555, given in figure 5. Before a trigger pulse arrives to trigger the timebase the trigger input (emitter of T4) is held high by R30, so T4 is turned off and the collector voltage is at + 15 V. The trigger pulse grounds the emitter of T4, turning it on so that the collector voltage, and hence the voltage at the top end of R29, falls. This takes the voltage on the inverting input of comparator 2 of the 555 below the volt- age on the non-inverting input, so the output goes high, setting flip-flop 4 (fig- ure 5). T1 (figure 5) is turned off, removing the short across the timing capacitor (block A in figure 4a), which is now charged linearly by the constant current source T2. When the voltage on the timing capacitor exceeds 2/3 supply voltage the output of comparator 1 (figure 5) goes high, resetting the flip- flop, which turns on T1 and shorts out the timing capacitor. Figure 4b. Timebase range switch and timing capacitors. Figure 5. Internal circuit of the 555 timer. Figure 6. Circuit of the X and Y output amplifiers. Figure 7. Printed circuit board and compo- nent layout for the X and Y output ampli- 1-34 — elektof january 1977 elektorscope (2) To avoid current being robbed from the capacitor, which would spoil the ramp linearity, the ramp output is buffered by a high impedance amplifier compris- ing T5 and T6. The sweep output is taken from the collector of T6. During the sweep the trigger input is inhibited by T3. This transistor is turned on when the ramp voltage exceeds its Vbe an d it shorts the base of T4 to ground for the duration of the sweep. This also facili- tates the free-run facility. When the trigger circuit is in the free-run mode, or in the auto mode with no input signal, the emitter of T4 is permanently grounded. At the start of each sweep T4 will be turned on and will trigger the 555. It will then be turned off by T3 until the 555 resets, when T3 will turn off and T4 will turn on again, thus retriggering the timebase. The flyback blanking pulse, which is obtained from pin 3 of the 555, is routed through the beam switching circuit where it is gated with the chop mode blanking pulses. Fine speed cali- bration of the timebase is provided by P2, which varies the current through T2, and X-position control is provided by P3, which applies a variable DC bias to one input of the X amplifier. Timebase range selection is performed by Sla (figure 4b), which switches in various values of timing capacitor. This figure corresponds to block A in fig- ure 4a. A second bank of this switch, Sib, is used to switch automatically from the ‘chop’ to ‘alternate’ channel switching mode. From 100 ms/cm to 1 ms/cm the ‘chop’ mode is used, and from 300 jus/cm upwards the ‘alternate’ mode is used. X and Y output amplifiers It will be convenient at this point to describe the X and Y output amplifiers, since an understanding of their oper- ation is necessary to understand the design philosophy behind the channel- switching circuits. Each amplifier (figure 6) consists of a differential cascode amplifier, and the X and Y amplifiers are identical except for the x5 trace expand switch (S8) of the X amplifier. As an example, the Y-amplifier proper consists of a differential amplifier (T3/T4) with a current source (T5) in the common emitter connection. This configuration, known as a ‘long-tailed pair’, can handle relatively large input voltage swings. The ‘cascode’ transistors T 1 and T2 serve as output buffers. They are used in a grounded base configur- ation, at a base voltage of 15 V. This arrangement has two advantages: the differential amplifier proper operates at a reasonably low collector voltage, so high-gain transistors can be used, and there is very little internal feedback from output to input, so there are prac- tically no stability problems. The gains of the amplifiers may be adjusted by P 1 and P2 and by P3 in the x5 position. Both amplifiers operate from the +150 V H.T. rail, and their outputs are connected direct to the X and Y plates. To avoid problems due to loading of the outputs by long leads the X and Y amplifier p.c. board is mounted directly behind the CRT base. The layout of this board is given in figure 7. Electronic switches The signal switching arrangements in the Elektorscope are fairly complicated. Firstly, the (differential) outputs of the Y preamplifiers must be routed to the inputs of the Y output amplifier, either one at a time when only Y1 or Y 2 is selected, or alternately at high speed in the Y1/Y2 chopped or alternate modes. Secondly, switching must be included to transpose the outputs of the Y2 pre- amplifier to invert the trace. Finally to switch between the normal and X-Y modes it must be possible to route either the timebase outputs, or the out- put of the Y 1 preamplifier, to the inputs of the X amplifier. To avoid problems that may be caused by long lengths of lead, such as capaci- tive loading, instability, hum pickup etc. the switches cannot be mounted on the front panel but must be located so that the signal path between the Y preamps and the output amplifiers is as short as possible. This means mounting the switches at the back of the Y preamp boards. Commercial oscilloscopes overcome the problem of controlling such switches by ingenious mechanical linkages such as extension spindles, rods, Bowden cables 9 r-J ©15V -w-o^ — H— f Figure 8. Showing the principle of the elec- tronic switch. Figure 9. The electronic switch may be extended to two or more channels, still con- trolled by a single transistor. Figure 10. Circuit of one Y preamplifier. Photo 1. The completed output amplifier Table 1. Test point voltages for the Y preamp. elektorscope (2) eiektor january will always be about 0.6 V higher due to the forward voltage drop across Dl. The output will follow the signal at the junction of Dl and D2, but will always be 0.6 V lower due to the forward voltage of D2, i.e. it will be equal to the input voltage. Provided Dl and D2 have approxi- mately the same characteristics, any distortion introduced due to variations and the like, but such solutions are not suitable for the amateur, who generally likes the mechanical arrangements to be simple. In addition the Y1/Y2 channel switches must be completely electronic as they have to operate at the chopping frequency of 50 kHz. For these reasons it was decided to make all channel switching in the Elektorscope com- pletely electronic. This has the ad- ditional advantage that all these compli- cated functions can be controlled by single pole switches. The principle of the electronic switch is shown in figure 8. When T1 is turned off Dl will always be forward biased provided the input voltage does not exceed the H.T. voltage (+15 V). The voltage at the junction of Dl and D2 will thus follow the input signal, but 1-36 - elektor january 1977 alektorscopa (2) in the forward voltage of D1 with variations in the current through it will be cancelled by similar variations in the voltage drop of D2, provided the input and output impedances are similar. When T1 is turned on by feeding current into its base the voltage at the junction of Dl, D2 and D3 is held to just above the forward voltage drop of D3 (slightly greater due to the saturation voltage of Tl) provided the input volt- age remains above 0 V. Again about 0.6 V is dropped across D2, so the out- put voltage is approximately 0 V. When the output of the switch is connected in parallel with other switches then D2 will be cut off by any signals present on their outputs, provided the voltage is greater than 0 V. When the electronic switch is in the ‘off position therefore, it will not load the output of any other switch that is on. If a switch is to control more than one signal channel then it is necessary to duplicate only the resistor/diode net- work, not the transistor, as shown in figure 9. D3 and D3’ isolate the two channels from each other. Y preamplifiers The circuit of one Y preamplifier is given in figure 10. It will be seen that, with the exception of the trigger output taken from emitter follower T5, the cir- cuit is completely symmetrical. This improves temperature stability and also provides the differential outputs necess- ary to drive the output amplifiers. The input stage consists of a dual- FET T10 connected in a differential source-follower configuration. Use of a dual-FET means that the two devices are matched and are also in close thermal contact so that their character- istics will track together with tempera- ture changes, thus minimising drift of the DC conditions within the amplifier. It is possible to use two E300 FET’s mounted in a common cooling clip, but it may be necessary to experiment with the value of R21 to achieve a balanced condition at the outputs of the preamp. The next two stages of the preamp are also differential, comprising transistor pairs T1/T3 and T2/T4. With the front panel gain control P2 in the ‘cal’ pos- ition the gain of the preamp may be varied between about 10 and 50 by P3. Provision has been made on the p.c. board layout for a compensation circuit P4/C21 to extend the h.f. response of the preamplifier. However, this is included only for the benefit of the serious experimenter and these compo- nents are not included in the parts list. PI is the Y-position control. To enable the preamplifier to drive the output amplifier and the inherent capacitances in the electronic switches without undue loss of bandwidth, emitter followers T6 and T7 are pro- vided. The electronic switches consist of T8, T9 and their associated diode networks D3 to D14. In the case of the Y1 1-38 — elektor january 1977 elektorscope (2) 11 preamp JT9 controls the switching of the Y and Y outputs to the inputs of the Y output amplifier, while T8 controls the switching of the outputs to the inputs of the X amplifier for the X-Y mode. In the case of the Y2 preamp T9 again controls switching to the Y output amplifier, but T8 switches the outputs of the Y preamp to the Y amplifier inputs in a transposed manner for the trace invert mode (i.e. Y output to Y input and Y output to Y input) Y attenuator When connected to the Y amplifier and correctly calibrated the Y preamplifier has a basic sensitivity of 10 mV/cm. To display larger input signals without ex- ceeding the screen limits some form of attenuator must be incorporated. This is connected behind S7 in figure 1 0 and its circuit is given in figure 1 1 . The input resistance of the Y preamp is 1 M, and the input attenuator consists basically of resistive potential dividers designed to maintain a constant input resistance of 1 M while giving input sensitivities of 30 mV, 100 mV and so on. However, the 1 M input resistance is also shunted by a capacitance of a few pF due to the gate capacitance of the FET, and if this were left uncompen- sated it would form a low-pass filter with the series arm of the attenuator, leading to an early roll-off in the fre- quency response. To avoid this the attenuator resistors are parallelled by capacitors. When the trimmer capacitors are correctly adjusted the reactances of the capacitors (taking into account the preamp input capacitance) will be in the same ratios as the attenuator resistors, so the attenuation factor will remain constant whatever the frequency. How- ever the input impedance will reduce with increasing frequency due to the falling reactance of the capacitors. It will be noted that the lower resistors in the attenuator sections are shunted with in some cases fairly large values of capacitance, and at first sight this may seem strange: it is the undesirable effects of the FET shunt capacitance for which we are trying to compensate, so why make it deliberately worse? The reasons for this are twofold. Firstly, this arrangement gives a fairly constant input capacitance of around 30 pF. This is necessary if the oscilloscope is to be used with high impedance probes, since these are designed to work into an input resistance of 1 M in parallel with between 20 and 40 pF. Secondly, the use of shunt capacitors avoids the necessity for impossibly small trimmer capacitors. For example if R2 were shunted simply by the 5 pF or so of the FET then R 1 would have to be shunted by a trimmer whose reactance was ^330 ~ t * mes as muc h, or t0 P ut another way, whose capacitance was .0003 times the FET capacitance — about .0015 pF, which is ridiculous. The Y attenuators are mounted on the same p.c. board as the preamplifiers, the Figure 11. The compensated Y input attenu- ator. layout of which is given in figure 12. For screening and stability an earth plane is provided on the upper side of this p.c. board, and care should be taken not to let the bodies of components short to this ground plane. This applies particularly to unsleeved electrolytic capacitors, to capacitors with metal end caps (e.g. Siemens MKM types) and to resistors, which may have only a thin paint film covering the end caps. The best plan is to place a thin strip of card beneath each component while soldering to stand it off from the board, or alternatively to give the top of the board a good coating of lacquer or insulating varnish. NOTE All component lists will be given in the final part of this article. M quadi-complimentary complemented elektor january 1977 — 1-39 Several readers have sent us comments and queries concerning the article Quadi- Complimentary (Elektor, December 1975). Some further explanation appears to be in order. The original article dealt with the prin- ciples involved in Quad’s ‘Current Dumping Amplifier’ type 405. The majority of the readers request us to investigate the possibility of using this principle in a new amplifier design. Before going into details, it should be plainly stated that this will be practi- cally impossible if the final design is to be a reliable ‘home construction’ project in the Equa, Edwin, Equin tradition. This is why we are still looking for alternatives; see ‘Ejektor’ in last month’s The ‘current dumping’ principle is unsuitable for a home construction project; it should only be considered in mass-produced commercial amplifier designs. Why? This can be made clear as fol- lows. In the case of ‘current dumping’ it is of prime importance that four im- pedances Zl, Z2, Z3 and Z4 should be very accurately defined. And that is the main difficulty! Figure 1 shows the overall block diagram; the conditions for Zl . . . Z4 are given in figure 2. The impedances Zl and Z3 of figure 1 are shown as resistors R1 and R3 in figure 2. For various reasons, Z4 be- comes a self inductance L4. Unfortu- nately, in practice this will be a coil rather than a pure self inductance, so there will be a resistance R4 in series with the coil. This, in turn, means that Z2 must consist of a capacitor C2 and a resistor R2 connected in parallel, in order to achieve the required bridge balance. The bridge is in balance if: Z x • Z 3 = Z 2 • Z 4 *) The transformation from figure 1 to figure 2 is defined by: Z, =R, Z 3 = R 3 z - R * 2 1 +jwRjCj Z 4 = R 4 + jojL 4 In this case there are two conditions for bridge balance: Ri R 3 = R 2 R 4 , and L* = Rj R 3 C 2 = R 2 R 4 C 2 , or In the theoretical case where R 4 is zero, R 2 becomes infinitely large and can be omitted. In actual practice, however, two align- ments are needed, both equally diffi- cult. In an experimental circuit it proved necessary to use a preset poten- tiometer for R 2 and a trimmer for C 2 . The slightest deviation from bridge balance resulted in ugly cross-over phenomena. A simple quiescent current adjustment as used in the Equin, for example, is a lot easier and gives better guaranteed results For a mass-produced version, the situation is somewhat different. There the specifications of the passive components Z 3 Z 4 are a dead fix, so it is possible to eliminate the alignment procedure and still get (very) good results. 14 Literature: Quadi Complimentary, Elektor 12, December 1975, p. 1220. •Several readers have remarked that the assumption of an infinitely large open-loop gain also implies an infinitely large feedback factor. At first sight it would appear that this is what eliminates the non-linearity of the output stage. However, this assumption is incorrect. The balance condition for a finite Z 2 Z4 = Z,Z 3 + (Zl + Z2) Z3 This is where 'current dumping' differs fundamentally from the Edwin principle: in the latter case the highest possible Ao, c.q. feedback factor, is indeed required. 1-40 — elektor january 1977 ii-fet opamps UMMS LilXilitE Interesting results can often be obtained by using several different types of active device in one circuit. Examples of this type/bf 'hybrid' circuit are designs incorporatingitoth transistors and valves (or 'tubes' Am. or 'bottles' SI.) or TTL and CMOS, or bipolar transistors and FETs. A relatively new example of this is the Bi-FET opamp. It consists of J FETs and bipolar tran- sistors on the same chip, and this could well give interesting results at a reasonable price. The manufacturer describes the new opamps as ‘monolithic JFET-input op- erational amplifiers incorporating standard bipolar transistors on the same chip, manufactured using ion-implan- tation techniques’. Furthermore, ‘these amplifiers feature low input bias and offset currents, low offset voltage and offset voltage drift, offset adjust which does not degrade drift or common-mode rejection, high slew rate, wide band- width, extremely fast settling time, low voltage and current noise and a low 1/f noise comer’. Since they are also supposed to be rugged and relatively cheap, it would appear that some further investigation is called for . . . The simplified circuit diagram of the new opamps (National Semiconductor types LF355, LF356 and LF357) is shown in figure 1 . The ion-implantation manufacturing technology makes for well-matched FETs in the input stage, so that the input offset voltage can be very small. The input impedance is 10“ and the input bias current is only 30 pA. As the technical specifi- cations show (table 1) the input noise figures are also exceptionally low. The second stage is a ‘long-tailed pair’ of bipolar transistors. This stage provides most of the open-loop gain (106 dB, or 200,000 x !). The output stage is a rather unusual hybrid quasi-complementary circuit. The basic circuit is shown separately in figure 2. It is short-circuit protected and it can withstand capacitive loads up to 10 nF without danger of instability. These BI-FET opamps are eminently suitable for use in precision high-speed integrators; fast D/A and A/D con- verters; high impedance buffers; wide- band, low-noise, low-drift amplifiers; logarithmic amplifiers; sample and hold circuits; etc. They can also be used, of course, in conventional opamp circuits. One of the few conventional things about these opamps is the pinning: they are pin-compatible with the well- known 741, as shown in figure 3. The only thing to watch is that the offset control is connected to the positive sup- ply - not the negative supply as with the 741. Applications. A basic wide-band amplifier circuit using the LF357 is shown in figure 4. If required, an offset control can be in- cluded as shown in figure 3. The gain is x 10, the p-p output voltage swing is 20 V, the power bandwidth is 500 kHz and the distortion is less than 1%. A more sophisticated circuit, using the LF356, is shown in figure 5. The parasitic input capacitance Cl consists of the input capacitance of the opamp (3pF or less) and any additional capaci- tance introduced by the printed circuit board of other wiring. It can be com- pensated for by including C2, where In this case the power bandwidth is | approximately 240 kHz at the same I distortion level ( 1 %). A more interesting application is shown in figure 6. This is a sharp notch filter; provided the components in the twin-T network are sufficiently well matched a quality factor Q of more than 100 can be obtained. As an example , if R = 2R1 = 10M and C = ^ = 300 p the centre frequency of the notch will be approximately 100 Hz and the ‘depth’ will be —55 dB. Final notes The National Semiconductor opamps described here are the first of a new generation. NS have already announced a FET-input 741, the LF 13741. This has an input bias current of 200 pA, but in all other aspects (bandwidth, gain and rise-time) it has the same specifications as the 741. Quite recently they introduced the LF352 series; these are instrumentation amplifiers with high gain linearity for low-level input signals and a high common-mode rejection ratio. They have also announced that the LF355, I 356 and 357 are now to be made avail- able in a much cheaper 8-pin mini-DIP version. Furthermore, the opamps described i here are already being supplied by Fair- child and Texas Instruments as second sources. This is often considered the ‘acid test’ of a new technology: how long does it take for other manufac- turers to jump on the bandwagon ] (National Semiconductor application note). Figure 1. Simplified circuit diegram of the BI-FET opamps. Figure 2. Basic circuit of the output stage. Figure 3. Pinning of the LF355, LF356 and LF357. They are pin-compatible with the 'traditional' 741, but the offset compensation is connected to the positive supply. Figure 4. A basic wide-band amplifier circuit. Figure 5. A more sophisticated circuit, includ- ing compensation for parasitic capacitances. Figure 6. Circuit for a high-Q notch filter. 1-42 — elektor January 1977 ic audic LG GGGLG In this article two completely different types of high-power audio IC's are discussed. To avoid confusion the article is split into two parts. The first part deals with hybrid power-integrated circuits while the second section deals with the TDA 2020 IC. In a monolithic audio IC the entire cir- cuit is integrated onto a single chip or die. As this chip is extremely small the problems involved in conducting away the heat dissipated in the circuit are enormous. Higher power audio IC’s (greater than 20 W) generally use a hybrid form of construction whereby the output devices are on separate chips from the rest of the circuit, which may be integrated using monolithic or thick film techniques. The individual parts of the circuit are bonded to a large metal substrate and, after interconnections have been made, the whole assembly is encapsulated in resin or silicone rubber. The metal substrate provides good thermal contact with a heatsink, though the package is, of course, somewhat more bulky than a 14 pin DIL IC! There are many hybrid audio IC’s currently available, with power outputs up to more than 1 00 W, but of particu- lar interest are a family of IC’s manufac- tured by ITT, Skiltronics and Sanyo. These are of special interest a) because they are available with out- put powers from 20 to 40 W in an identical package, so the same circuit layout can be used for all versions. b) because they are obtainable from three different manufacturers, so they should be easier to obtain. The maximum ratings and principal electrical characteristics of this family are given in table 1. Note that the Skiltronics type numbers are prefixed SPH, while the ITT and Sanyo type numbers are prefixed STK. Figure 1 shows the external components required by the amplifier. The number of external components may seem rather large, but of course many of these are electrolytic capacitors, which are difficult to incorporate into a small package, and which might be dam- aged by the high temperatures that can occur within the IC. It will be noted that these circuits operate from a symmetrical supply (figure 2) with the loudspeaker direct coupled to the out- put. Since the IC’s are used as function building blocks or ‘black boxes’ the internal circuitry will not be discussed in detail. However, it should be stated that the output stage is of the quasi- complementary type, and the input stage is a differential amplifier of a fairly common configuration. Precautions Provided a few simple precautions are observed few problems should be encountered when using these IC’s. These are as follows: 1 . Both the positive and negative supply voltages must be applied to the IC simultaneously, since if only one voltage is applied a large d.c. offset will appear at the output, causing a large current to flow through the loudspeaker that could damage both loudspeaker and IC. Even a fuse in series with the loudspeaker will not protect the IC in these circum- stances. ic audio Figure 1. Application circuit for the 022, 025 032 and 036 IC's. Figure 2. The Hybrid IC amplifiers require only a simple power supply. Figure 3. Printed circuit board and com- ponent layout for the 022, 025, 032 and 036 IC's (EPS 9439a). To minimise the d.c. current that flows through the loudspeaker (and hence the ‘thump’) when switching the amplifier on and off it is essential that the rise- times of both supply rails should be similar. Since the supply need only be a transformer, bridge rectifier and two reservoir capacitors this simply means that the reservoir capacitors should be of the same value. This will also ensure that the two supply voltages are as nearly the same as possible, which will minimise the d.c. offset voltage of the amplifier. This will normally be no more than ± 50 mV. 2. The second precaution is to ensure that the heatsink is of an adequate size, since these ICs do not incorporate thermal protection. The thermal resist- ance of suitable heatsinks is given in table 1 . These are adequate even if the amplifier is driven for long periods at full output. For normal domestic use smaller heatsinks may be adequate. When mounting the IC on a heatsink it is essential that the spacing of the fixing holes should be as accurate as possible, otherwise the heatsink or baseplate of the IC may become dis- torted, leading to poor thermal contact. Some silicone heatsink grease could be used to improve thermal contact. 3. It might be thought that fusing the positive and negative supply rails would be a suitable method of protection, but this is not the case since failure of a fuse in one rail would cause the fault con- dition discussed earlier, i.e. a large d.c. offset on the output. The only fuse should be in the primary of the mains transformer, and this is intended to protect the power supply rather than the amplifier. Construction A printed circuit board and component layout suitable for use with IC’s type 025, 032 and 036 is given in figure 3. The input resistor R1 forms an attenu- ator with the input impedance of the 1C and should be chosen to provide the required input sensitivity/input im- pedance, taking into account the gain of the IC. For example, suppose the STK032 is required to provide 25 W into an 8 ft load with an input sensitivity of 1 V. 25 W into 8 S2 means an rms output voltage of approximately 14 V. The gain of the IC is about 30 dB or 31 times, so the voltage required at the input is 14/31 or 0.45 V. Since the required input sensitivity is 1 V this means that 0.55 V must be dropped across R1 so Rl is x Zj n or about 33 k. The input impedance of the amplifier (with Rl) is thus 60 k. R1 and Cl together form a lowpass filter to protect the IC against high slew- rates (which gives rise to transient intermodulation distortion) by limiting the risetime of the input signal. If Rl is changed then Cl should also be changed to maintain the same break- — elektor January 1977 Table 1 . Maximum ratings, electrical characteristics and applicatic audio IC's for use with symmetrical power supply Figure 5. Circuit for a 20 W amplifier using i single supply rail. Table 1. Pertinent data for the 022, 025, 032 and 036 IC's. Table 2. Absolute TDA2020. Table 3. Electrical characteristics of the TDA2020. maximum ratings TDA2020. Supply voltage Input voltage Differential input Output peak current (internally limited) Power dissipation at Tease <75°C Storage and junction temperature —40 SPH 022 SPH 025 SPH 032 Skiltronics, ITT. Sanyo - STK 025 - STK 022 STK 025G STK 032 Maximum supply voltages ±25 V ±29 V ±32 V Nominal supply voltages ±19 V ±22 V ±24 V quiescent current 30-50 mA 4 ohm Minimum output power into g Qhm 15 W 20 W 25 W Maximum case temperature 85° C Maximum output short-circuit two seconds under full drive duration recommended mains transformer 2x 15 V 75 VA 2 x 17 V 90 VA 2x 18 V 100 V A speaker fuse, if required 5 . . .6 A, fast blow voltage gain (dB) 33 dB 30.5 dB 30.5 dB input impedance 27 k recommended thermal resistance of heatsink 2° C/W 1.7° C/W 1.5° C/W 1 bridge rectifier 40 V 2 A 40V/3.2A Reservoir capacitor >2200 p >3300 n >3300 p elektor january 1977 - 1-45 Table 3 Electrical characteristics TDA2020 (V s = ±17V, T a mb = 25°C unless otherwise specified) Parameter Test conditions Min. Typ. Max. Unit V s Supply voltage t5 ±22 V Id Quiescent drain V 5 - ±22V 60 mA lb Bias current V S -±17V 0.15 pA Vi (off) Input offset voltage 5 mV li(off) Input offset 0.05 v o(off) Output offset voltage 10 100 Po Output power d = 1% G v = 30 dB Tease <70°C f =40 to 15.000 Hz V$= ±17V R L = 4 n V s =i18V R[_= 4 n V S =±18V R L =8n 15 18.5 20 16.5 W W d = 10% G v = 30 dB Tease ^ 70°C f =1 kHz V s = ±1 7V R|_= 4 n V S =±18V R L = 8 n 24 20 W Vj Input sensitivity G v = 30 dB f = 1 kHz P 0 = 15W V s = ±17V R|_ = 4 n V S =±18V R|_=8n 260 380 mV B Frequency response (—3 dB] R|_= 4 Cl C4 = 68 pF 10 to 160,000 Hz d Distortion P 0 = 150 mW to 15W R|_= 4 SJ G v = 30 dB Tease ^70°C f = 1 kHz f =40 to 15,000 Hz 0.2 0.3 1 % % P 0 = 150 mW to 15W V S =±18V Rl= 8 n G v = 30 dB T case <70°C f = 1 kHz f =40 to 15.000 Hz 0.1 0.25 % % 5 point for the filter. As a rule-of-thumb, if R 1 is lk then Cl should be 3n3, and if R1 is increased then Cl should be descreased by the same factor. For instance, if R1 is increased to 10 k (multiplied by 10) then Cl should be divided by 10 (reduced to 330 p). When making these calculations the output impedance of the source feeding the amplifier (e.g. a preamp) must be taken into account. For example, if the preamp has an output impedance of 5 k and R1 is 1 k then the effective value of R1 is 6 k, so the nominal value for Cl should be divided by 6. In practice, of course, the value of Cl is not so critical. The value shown in figure 1 (ln5) can be used for any value of R1 between 1 k and 5k6, for instance. Furthermore, the minimum value is 330 p. Distortion Finally, figure 4 includes graphs of one important parameter not shown in table 1 - distortion. Figures 4a and 4b show harmonic distortion versus power output at three frequencies, into 8 Si and 4 Si loads, for the 025 IC, while figure 4c shows intermodulation distor- tion versus output power into 8 and 4 Si loads. Twenty twenty The next type of audio IC to be dis- cussed is the SGS TDA2020. It is a monolithic audio IC that can achieve an output power of (typically) twenty watts continuous. Furthermore, cross- over and harmonic distortion are low so that the IC is truly ‘hi-fi’. The chip incorporates safe area limiting to keep the output transistor dissipation within acceptable limits, together with a thermal shutdown system to protect the whole IC against overheating. The absolute maximum ratings of the TDA2020, which should never be exceeded, are given in table 2. The electrical characteristics are given in table 3. The TDA2020 may be used with either a symmetrical (±) power supply, in which case the loudspeaker can be direct-coupled to the output stage, or with a single power supply rail, in which case an output coupling capacitor is needed. Although a direct-coupled output theor- etically saves components, it has the disadvantage that the loudspeaker is unprotected against d.c. fault conditions in the amplifier, and the supposed saving is more than outweighed by the need to provide loudspeaker protection circuits. For this reason the practical circuit is based on an amplifier with a single power rail, as shown in figure 5. In this circuit, 100% d.c. feedback is provided via R5, and the non-inverting input is biased to half the supply volt- age by R1 and R3. This means that the quiescent output voltage of the ampli- fier is also half the supply voltage. The gain of this circuit is about 30 dB. Maximum output power of 1 6 W into 8 S2 is thus achieved with an input volt- age of 360 mV rms, or with a 4 load 20 W may be obtained with a 285 mV input. With an 8 £2 load, distortion at 1 kHz is less then 0.1% for output powers from 150 mW to 15 W. Distortion into a 4 £2 load is approximately twice this. Figure 6 shows a printed circuit board and component layout for this particu- lar application circuit. A special spacer (supplied with the TDA2020) is Figure 6. Printed circuit board and com- ponent layout for a 20 W amplifier using the TDA2020 (EPS 9144). Figure 7. A suggested layout for an 'all IC' stereo amplifier using the TDA2020 and the TCA 730/740 control amplifier. mounted beneath the device before soldering it into the board. This supports the heatsink, which is mounted on top of the IC and is secured by bolts passed through the spacer. Good thermal con- tact between the IC and the heatsink is ensured by a copper slug which is integral with the IC package. Some heat- sink compound should be smeared on the IC and the back of the heatsink to improve this contact still further. The exact shape of the heatsink is unimportant, provided it fits the board. However, if the IC is to be run at full power for long periods of time then the thermal resistance of the heatsink should not be less than 2°C per watt. In normal domestic use heatsinks of up to 8 C per watt may be acceptable. It should be noted that the copper slug (and hence the heatsink) is internally connected to pin 5, so the heatsink is at ground potential. However, if a sym- metrical supply is used, the negative supply potential is present on the heat- sink! Figure 7 shows a suggested layout for a simple ‘all IC’ stereo amplifier using two rDA2020’s and the TCA730/740 con- trol amplifier described in Elektor 8. If a disc input were required then the con- trol amplifier could be preceded by the IC disc preamplifier featured in Elektor 3. The power supply to the TDA2020’s may be unstabilized and is derived simply from a transformer, bridge rectifier and reservoir capacitor. The 1 5 V stabilized supply to the control amplifier may be obtained either from a suitable IC voltage regulator, or from a simple zener stabilizer. If the IC disc preamp is also used this, of course, has a built-in voltage regulator. 14 1-48 - elektor january 1977 the BF 494 a case in point trUEl Editor's lament: Oh why, tell me, why Can our readership's eye More clearly descry Than an author, or I, Any slip, imperfection, Or faulty connection? The BF 494 is an extremely useful HF transistor, and it is used in many Elektor circuits for this reason. The specifications are quite good and the price is quite reasonable, which means that it can be used as a kind of ‘Univerai HF Transistor’. Applications range from found that these two reliable sources had different ideas about the BF 494 . . A few ‘phone calls to the two parties concerned taught us that, in this case, the Philips data are accurate. Pro- Electron, for once, appears to have slipped up. Figure 2 therefore gives the official pinning for the BF 494. As a final note we would like to remark that we were highly impressed by the rapid and energetic action undertaken by the ‘Association Internationale Pro- Electron’ to locate and remedy the mis- take. We will continue to rely on their data handbooks in the future, albeit perhaps not quite so blindly ... M Figure 1. The BF 494 according to Pro- Electron. Figure 2. The BF 494 according to Philips. Note the interchanged collector and emitter 116 (TO- 92) 9784 1 1 oscillator and mixer stages in AM and FM receivers to HF and IF amplifier stages. However, there is a problem. As several observant readers have pointed out, the pinning shown in the transistor list in El 7, p.947, is at variance with the pin- ning used on the Elektor printed circuit boards. As we mentioned in the ‘Missing Link’ last month, the information in the transistor list was incorrect; this has been corrected in the more recent lists. The reason for the slip is perhaps interesting. We took extreme care to keep the list in exact accordance with the ‘Standard Handbook’ issued by Pro-Electron. When the reader’s en- quiries started coming in, we compared the Pro-Electron data (figure 1 ) with the pinning shown in the Philips data hand- book (figure 2). To our surprise we 7404+ (7405*. 7406*. elektor january 1977 1-59 1-60 — elektor january 1977 tup-tun-dug-dus TUP ▼UP Tun run DUE DUE DUS Wherever possible in Elektor circuits, transis- tors and diodes are simply marked 'TUP* (Transistor, Universal PNP), 'TUN' (Transistor, Universal NPN), ‘DUG' (Diode, Universal Ger- manium) or 'DUS' (Diode, Universal Silicon). This indicates that a large group of similar devices can be used, provided they meet the minimum specifications listed in tables la and 1b. For further information, see the article ‘TUP- TUN-DUG-DUS* in Elektor 1, p. 9. BC 253 BC 261 BC 262 BC 263 BC 307 BC 308 BC 309 BC 320 BC 321 BC 322 BC 350 BC 351 TUP « , i DUE BC 352 BC 415 BC 416 BC 417 BC 418 BC 419 BC 512 BC 513 BC 514 BC 557 BC 558 BC 559 NPN PNP | BC 107 BC 108 BC 109 CD CD CD o o o V C e 0 45 V 45 V 20 V 25 V 20 V 20 V Veb 0 6 V 5 V 5 V 5 V 5 V 5 V 100 mA 100 mA 100 mA 100 mA 100mA 50 mA 300 mW 300 mW 300 mW 300 mW 300 mW 300 mW »T 150 MHz 130 MHz 150 MHz 130 MHz 150 MHz 130 MHz F 10 dB 10 dB 10 dB 10 dB 4 dB 4 dB The letters after the type number denote the current gain: a' (0. hf e ) = 125-260 a' = 240-500 a' = 450-900. Table 6. Various equivalents for the BC107, -108, . . . families. The data are those given by the Pro-Electron standard; individual manu- facturers will sometimes give better specifi- cations for their own products. NPN PNP Case Remarks BC 107 BC 108 BC 109 BC 177 BC 178 BC 179 •0 BC 147 BC 148 BC 149 BC 157 BC 158 BC 159 ■g 250 mW BC 207 BC 208 BC 209 BC 204 BC 205 BC 206 ■© BC 237 BC 238 BC 239 BC 307 BC 308 BC 309 •Q BC 317 BC 318 BC 319 BC 320 BC 321 BC 322 GE Icmax = 150 mA BC 347 BC 348 BC 349 BC 350 BC 351 BC 352 (D! BC 407 BC 408 BC 409 BC 417 BC 418 BC 419 ■a Pmax = 250 mW BC 547 BC 548 BC 549 BC 557 BC 558 BC 559 <3 TOO mW BC 167 BC 168 BC 169 BC 257 BC 258 BC 259 3! 169/259 50 mA BC 171 BC 172 BC 173 BC 251 BC 252 BC 253 ■<3 251 .. . 253 BC 182 BC 183 BC 184 BC 212 BC 213 BC 214 ■<3 200 mA BC 582 BC 583 BC 584 BC 512 BC 513 BC 514 •3 To mA BC 414 BC 414 BC 414 BC 416 BC 416 BC 416 ■3 low noise BC 413 BC 413 BC 415 BC 41 5 ■3 low noise BC 382 BC 383 BC 384 3 BC 437 BC 438 BC 439 B m 220 mW BC 467 BC 468 BC 469 01 Pmax = 220 mW BC 261 BC 262 BC 263 ■0 Oder a sUoscriplicn belekbr or 1977 new ctVfo orbes We regret that the cover price of ELEKTOR has been increased for 1977. BUT YOU CAN STILL subscribe to ELEKTOR for 1977 at £ 6.25*, so make sure you receive your copy regularly and in ADVANCE. Use the reply paid coupon, no stamp needed, and order your subscription now! * £ 6.25 UK and overseas surface mail. Look through this issue of ELEKTOR and you will find that ELEKTOR is not just one more Electronics magazine ELEKTOR is different ultterent — in its presentation — in its style and content — in its use of new techniques and components in practical circuits PLUS - An unequalled Printed Circuit Board Service — Technical queries service — and if you subscribe an additional circuit design in the mailing wrapper ELEKTOR is the magazine for your profession, study or hobby. ELEKTOR avoids long-winded theorectical discussions and the like. Instead, it gives practical applications. elekbr binders The dark green binder collects your loose copies of ELEKTOR into one hai volume. Each copy is fastened simply and firmly with a metal pin, and any copy can ea be removed later without disturbing the others. Matching self-adhesive labels are provided to show the year. ELEKTOR binders can be ordered from our Canterbury office using the form O' Price: £ 1.85 (incl. VAT and p&p). r? a i 0r new pn S 2 “° Another addition bon _ h '9h t “derated build - hi f„ n d suite specificat/o, '' yourself £ r fascinai ' « the c ■ncludinc quality ' . ''flht , "9 h power lahl_ The new Maplin Catalogue is no ordinary catalogue... |l details of a re ®'' v n ' i °^'?don system-E**' \l effective electronic a ^ , Q secon ds to 1\ reliable ve’ ta es « , h ould occun i rmei running- 1 _ ^ sssggSSgSsr 1 Catalogue includes a very wide range of components: hundreds of different capacitors; resistors; transistors; I.C.'s; diodes; wires and cables; discotheque equipment; organ components; musical effects units; microphones; turntables; cartridges; styli; test equipment; boxes and instrument cases; knobs, plugs and sockets; audio leads; switches; loudspeakers; books; tools - AND MANY MANY MORE. ^msssmsssr ,1 name address 1 Our bi-monthly newsletter keeps you up to date with latest guaranteed prices - our latest special offers Ithey save you 1 poundsl - details of new projects and new lines. Send 30p ■ for the next six issues (5p discount voucher with each copy). 1 delay in completion of larger warehouse, catalogue will be delayed by up to four weeks - so there's still time to order before publication and get your pack of ten super special offer coupons, giving big discounts on ten different popular items YOU COULD SAVE POUNDS! - SO DON'T DELAY - FILL IN AND POST COUPON NOW! mopun ELECTRONIC SUPPLIES P.0. BOX 3, RAYLEIGH, ESSEX SS6 8LR Shop: 284, London Road, Westcliff-on-Sea, Essex (Closed on Monday) Telephone: Southend (0702) 44101