up-to-date electronics for lab and leisure ELBHTOr ST 108 87- April 1975 35p elektor april 1975 - 405 412 417 418 422 423 427 427 428 selektor Under this heading elektor mill be publishing selected items about nem technologies and developments. tap preamp (part 1) Elektor readers will by now be familiar with the TAP or Touch Activated Programmer. For reliability and ease of operation all the preamplifier functions are controlled by TAP's and mechanical switches and potentiometers are eliminated. pll systems It is the intention of this article to give an introduction to Phase- Locked- Loop (PLL) systems, without assuming any advanced mathematical knowledge on behalf of the reader, nor any familiarity with the subject. decimal to bed converter — J. Wittje fido Fido is a new electronic game in which an unfortunate dog is called by four masters at the same time. If one of the players succeeds in getting Fido into his kennel, the game is decided: Fido stays where he is. elektor services to readers time machine - H.U. Heinz Slowly-developing technological processes or natural events cannot be perceived because the eye is generally not able to distinguish the separate stages. Such events and processes can, however, be visualised by means of cinematographic time compression. marine diesel noise generator — J. Jacobs minidrum (part 2) The Minidrum described in the previous issue may, by the addition of various extra circuits, be extended to a comprehensive manual drum kit. Some new instruments, a three channel ruffle system and an automatic bass- drum are described in this article. compressor Compressors are now being used on an ever-increasing scale. They may be found in tape recorders, intercom systems and baby alarms, public address systems, discotheques and of course broadcast transmitters. A com- pressor supplements a manual volume control and allows a system to adjust itself to a wide range of input signals with little distortion. The design described here should find a wide range of applications with the electronics enthusiast. disc preamp A preamplifier-equaliser for magnetic pickup cartridges has to meet quite exacting requirements. The well-known two-transistor configuration, operating from a 12 ... 18 V supply, invariably falls short on gain and overdrive- margin - unless it is designed for a low nominal output voltage (about 30 mV). An alternative approach is to make use of a good integrated amplifier. The design described here, which meets all the requirements, employs a SN 76131 . An almost identical I.C. is the ftA 739. a/d converter The necessity to convert a voltage to a frequency such that the frequency is accurately proportional to the voltage is one which arises in many different electronic systems. Some digital voltmeters use this principle. The circuit described here is relatively simple, but nonetheless the absolute error is less than 1 .5 mV and the relative error less than 1 % over an input voltage range of 7 mV to 2.5 V. led displays In response to numerous requests from readers we publish this LED-display chart to enable constructors to find their way through the jungle of seven-segment display data. modulation systems (part 2) Th 410 — elektor april 1975 tap-preamp parti tap preamp Figure 1. Touch panels for the TAP's. The contact surfaces and legends are nickel plated with a black background. Figure 2. The circuit of the four-position TAP. Touching one of the input contacts causes the corresponding output to become 'V and all other outputs to become 'O'. Figure 3. Circuit to show the principle of an electronic 'make' contact. The LED indicates that the contact is 'closed'. Figure 4. Extension of the circuit of figure 3 to control two channels. In a previous issue (elektor no. 2), in the article entitled 'Sonant', a new design of audio preamplifier and control unit was discussed, which would complement the power amplifier/loudspeaker combination of the Sonant. This article describes the design and construction of such a 'Pre-sonant', which combines high performance with simplicity of operation. Figure 5. The make contact applied to a four- preset-level volume control. The values of R 15‘ R 22 determine the four preset volume Figure 6. The ele ctr onic 'break' contact. When a ‘1 * appears at input Q„ T^ and T 2 are cut off and the LED lights to show that the contact is 'open'. Elektor readers will by now be familiar with the TAP or Touch Activated Pro- grammer. For reliability and ease of operation all the preamplifier functions are controlled by TAP’s and mechanical switches and potentiometers are elimin- ated. This necessarily leads to some simplification of control functions, as such things as volume and tone control can now be implemented only in discrete steps. This is perhaps no bad thing, as the front panels of some modem ampli- fiers look like something from ‘Star Trek’ and one wonders if a training course is necessary to operate them. This design is, therefore, not suitable for the dedi- cated knob twiddler! Assuming that the recording engineer has done his job properly, many control functions may be removed from the front panel of the preamp and may be replaced by internal presets. This applies to balance and tone controls, which may be adjusted to suit room acoustics and personal taste, after which no further adjustment should be necessary. The number of control functions was thus reduced to the following: Input Selection: Disc, Radio, Tape, Auxiliary. Four preset levels. Four settings from mono to ‘extreme stereo’. Bass lift, ‘Presence’, Rat, Treble cut. It is hoped in a later article to include a touch station selector for radio. The layout of the touch panels is shown in figure 1. These are available from the Elektor Print Service. Four Position TAP All the controls mentioned above are based on the four-position TAP shown in figure 2, which is designed around an RCA COSMOS IC type CD4011AE, a quad two-input NAND gate. The circuit operates as follows: When the circuit is first switched on the output of one of the gates will set to ‘1’ and all the others are held at ‘0’ since a ‘1’ is applied to their inputs via the input Volume: Image Width: Tone: resistors connected to +Vj, and via the diodes from the output of the gate whose output is M’. Which output sets to ‘1’ on initial switch on is determined by the switching speed of the individual gates and the various resistor tolerances. Suppose now that input 1 is touched. Pin 1 of gate Ni is now held at ‘0’ by the skin resistance, the output therefore be- comes ‘1’. This ‘1’ is applied to the inputs of the other three gates via D4 , D 7 and Dio respectively. Since the other input of each of these gates is already at ‘1’ via the input resistors R 4 , R s , R 7 , R g , R 10 and R n the output of N 2 -N 4 becomes ‘O’. The logic level on the anodes of D! , D 2 and D 3 becomes ‘0’ and pin 2 of N! is held at ‘0’ by R 3 . Thus when input 1 is released the output of N ! re- mains at ‘ 1 ’. This explanation applies for all the other inputs. Only one output can be a ‘1’ at any time. The TAP is used to control two types of electronic switch, a make contact, as shown in figure 3 and a break contact as shown in figure 6. When a ‘1’ is applied to the Q x input in figure 3, Tj is turned on. Current flows through the LED and resistor into the base of T 2 , which is also turned on. The LED lights to indicate that this switch position is activated. The modifications necessary to switch two channels are shown in figure 4. Ti is now used to switch two transistors and the base resistors are doubled in value (within the limits of preferred resistor values) to keep the LED current the The Break Contact The circuit of figure 6 operates in an inverse manner to that of figure 4. When the Q x input is at ‘0’ Tj is turned off. However, T 2 and T 3 are turned on by current flowing into their bases via the LED, R 2 , R 3 and 114. The ‘contact’ is thus normally ‘closed’. When a ‘1’ is applied to the Q x input T 1 is turned on thus grounding the bases of T 2 and T 3 and turning them off. Current flows through the LED via R 2 and Tj so that it lights. As an example of the use of the make contact a four-setting volume control is shown in figure 5 . For the left channel R13 and R 15 -R 2 i comprise a potentio- meter, likewise R 14 and Ri 6 -Rm for the right channel. When one of the inputs Q1-Q4 is high then the corresponding transistorsT s /T 6 -Tii/T| 2 are turned on, grounding one end of the corresponding collector resistor R 15 /R 16 -R 21 /R 22 • The attenuation depends on the value of the resistor that is grounded and may be varied to suit personal taste. After attenu- ation the signal is fed into the base of T,3 (T 14 ) and the output is taken from the collector. This and the other control circuits will be discussed in greater detail in next month’s article. H pll systems behalf of the reader, nor any familiarity with the subject. The need for such an introduction stems from the ever-increasing use of PLL circuits in consumer electronics and from the increasing complexity of these circuits, which is threatening to make new developments in this field incomprehensible to many electronics enthusiasts. The article also deals with Feedback PLL systems, which are in many ways superior to conventional PLL circuits. A simple receiver using the Feedback PLL principle will be described in a future issue. It is the intention of this article to give an introduction to Phase- Locked-Loop (PLL) systems, without assuming any advanced mathematical knowledge on A phase-locked-loop is a control system in which an electrical quantity is con- trolled by the phase difference between two signals. Figure 1 shows a block dia- gram of an arbitrary servo control sys- tem. Ax and Ay are quantities of the same form such as A.C. or D.C. potentials. These quantities are compared with one another in block C by, for example, multiplication or subtraction. The result of the comparison is processed in block C in such a way that quantity Ay is ad- justed. The form of processing deter- mines a number of the control charac- teristics such as the control time constant. Quantity Ay is readjusted in such a way that a state of equilibrium is reached at the output of C. Figure 2 is a block diagram of a PLL. In this case control is based on the phase difference between the input signal (1) and the signal (2) from a Voltage- Controlled Oscillator (VCO) so the con- tents of block must be able to recognize this difference. The VCO is controlled in such a way that a specific phase difference is main- tained between the output from the VCO and the input signal. The speed with which the PLL adjusts the VCO to follow any change in the input signal depends, in the first instance, on the characteristics of the low-pass filter LPF. When two signals are multiplied together, the product includes a component that is proportional to their phase difference and that can be filtered out from the other components. Block performs this multiplication. In practical circuits the input signal is multiplied by a square- wave output from the VCO, which means in effect that alternate half cycles of the VCO square wave multiply the input signal by +1 and —1. The waveforms in figure 3 should make it easier to under- stand the mode of operation. In figure 3a the input (represented as a sinusoid) is shown and below it a VCO square wave of the same frequency is repeated with phase relationships varying progressively from in-phase to 1 80 lead- ing (figures 3b, 3d, 3f, 3h and 3j). During the positive half-cycles of the VCO square wave (in any particular phase) the associated ‘product’ waveform (fig- ures 3c, 3e, 3g, 3i and 3k) is the same as the input sine wave of 3a. During the negative half-cycles of the square wave the sine wave of 3a is polarity-changed in the product waveform. This is equiv- alent to multiplying the two waveforms together. In the first product waveform (3c), which is associated with the in-phase square wave 3b, it will be seen that the product never becomes negative, in fact it is a full-wave rectified version of the sine wave. Its filtered D.C. value is thus unmistakably positive. When the square wave is leading by 45°, as in 3d, the product 3e clearly has a greater area above the line than below. Its mean D.C. level is therefore also positive, but less than 3c. When the square wave leads by 90°, as in 3f, the product 3g has equal areas above and below the line, so its D.C. value is zero. With leads greater than 90° the D.C. value of the product becomes negative, reaching a maximum (negative) value at +180° (3h to 3k). Summarising; the D.C. value of the prod- uct waveform varies from a maximum positive value when the square wave is in phase with the input signal, through zero when the square wave leads by 90°, to a maximum negative value when the square wave leads by 180°. Assume now that the input and VCO frequencies are precisely equal and that the PLL is locked in (ignoring, for the moment, how it got that way). The VCO square wave will be leading th input signal by 90° and the D.C. outpi of the phase comparator (multiplier) w be zero. Suppose now that the VCO fn quency tends to increase. The phase lea will become greater than 90° and tl D.C. output of the phase comparator w become negative. This will tend to t duce the VCO frequency and lock will l maintained with a slight increase ii phase lead. Conversely, if the VCO fn quency tends to decrease, the output ( the phase comparator wEl become pos tive, which will tend to increase the VO frequency. It can be shown that the input signal ci also lock to harmonics of the VCO fn quency, or the VCO to harmonics of th input signal (if the input signal is r sinusoidal as previously assumed). It j also possible to insert a frequency dividj between the VCO and the phase cob parator and by a combination of fn quency divider and harmonic locking tl ratio of VCO frequency to input frj quency can be made to assume peculij values such as 16/3 for example. Th] opens up intriguing possibilities for frj quency synthesis. The capture process Until now it has been assumed that tl PLL is locked in. It is now necessary I consider what happens when the circil is switched on and the VCO is out | lock, as it almost certainly will be. ' short answer is that the VCO hunts uni it finds a frequency and phase to whif it can lock. Some understanding of the capture pJ cess, as it is called, may fortunately j acquired without mathematics if tl behaviour of the circuit is examined I certain points in the loop and cert if assumptions are made. To assist in the explanation, assume fid that the connection between the Ll| output and the VCO input is broken. 1 VCO, deprived of a control voltage, * take up its free-running frequency whi may be assumed to be lower than tl input frequency. It has already beJ assumed, when discussing the locked! pll systems elektor april 1975 - 413 Figure 1. A control system consists of an infor- mation source Ax, a comparator circuit C, a processing circuit B and a controllable quan- FigureZ The elements of a PLL are: the phase comparator 0, the low-pass filter LPF, and the controllable oscillator VCO. Figure 3. Showing how the output of the phase comparator varies with the phase difference between the input signal and the VCO. Figure 4. Diagram to illustrate how the differ- ence frequency waveform changes during one cycle of the capture transient. condition, that the VCO frequency in- creases when the VCO control voltage goes positive and decreases when it goes negative. It may also be assumed that the LPF completely removes frequencies equal to the sum of the input and VCO I frequencies, that it passes D.C. with no | attenuation and that it passes the differ- [ ence frequency of the VCO and input [ signal with some attenuation, which de- I creases as the difference frequency de- creases (i.e. as the VCO frequency ap- proaches the input frequency). While the VCO is running free because of the supposed broken connection a differ- ence-frequency oscillation of constant amplitude appears at the LPF output. When the connection is re-made what next happens must be examined careful- ly. As pull-in has not yet taken place a difference frequency still exists and an oscillatory voltage is fed to the VCO con- trol input. Consider now one positive swing of the VCO control voltage from trough to crest (figure 4). The VCO control voltage is going positive, therefore the VCO fre- quency is increasing and the difference frequency is decreasing. Because of the decreasing difference frequency the attenuation of the difference frequency signal in the LPF will be progressively reduced and the overall swing of the VCO control voltage will have greater amplitude than with the VCO free- running. Figure 4a compares the positive- going swings under controlled and free- running conditions, starting from the same trough potential and time. The crest of the controlled swing is more positive and it occurs later because the difference frequency is decreasing. Figure 4b shows what happens during a negative (crest-to-trough) swing. Here the VCO control voltage is going negative and the VCO frequency is decreasing, so the difference frequency is increasing. Attenuation in the LPF is thus progress- ively increasing; overall amplitude is less than when free-running and the trough occurs sooner. Figure 4b is added onto 4a to show what will happen during one complete trough- 414 — elektor april 1975 to-trough cycle of the difference signal. The positive-going half cycle has a more positive peak than the free-running dif- ference signal. This ‘handicaps’ the nega- tive-going half signal and its reduced amplitude also helps to make the trough more positive than it would be in the free-running condition. Later cycles of the capture-transient, as it is called, cannot be compared with the free-running waveform, but they follow the same general pattern. Positive-going swings have increased amplitude while negative-going swings have reduced am- plitude. This results in both crests and troughs becoming progressively more positive whilst the time interval between them becomes longer. This means that the VCO frequency will also increase until a point is reached where one of these swings of the control voltage sweeps the VCO frequency through the input frequency. More swings may occur until the VCO has found the correct phase relationship before lock-in actually occurs. Applications of Phase Locked Loops A PLL provides two information out- puts. The VCO frequency, which is re- lated to the input frequency, and the VCO control voltage whose value de- pends on the phase difference between the input signal and the VCO output. If the desired information contained in the input signal is in the form of a fre- quency change (i.e. frequency modu- lation) then the PLL may be used as an FM detector. Its advantages over ratio detectors and coincidence detectors are: less distortion, better suppression of in- terference and the absence of LC circuits. PLL’s are also useful in frequency syn- thesis as figure 5 shows. In the example given in figure 5a the condition for lock- in is that fc/nv = f r and with a channel spacing of Af we have Af = f r . The fre- quencies delivered by the VCO are thus multiples of the reference frequency and it follows that the VCO frequency is itself determined by the division ratio n v . In many practical cases a variable-ratio divider will not be able to accept a high VCO frequency directly, so the VCO fre- quency is fed first to a stable fixed-ratio divider and from this to a stable adjust- able divider. With this procedure it is possible to divide down from a relatively high carrier frequency to a low channel- spacing frequency. This is useful in, for example, aircraft VHF equipment. In figure 5b an arrangement for fre- quency synthesis is shown in which delta pulses (needle pulses) recurring at the reference frequency from a crystal oscil- lator are fed into the phase comparator together with the VCO signal. As delta pulses contain the odd and even har- monics of the fundamental frequency the PLL can lock onto any harmonic. Construction of a PLL a. The VCO Requirements for the VCO depend, in the first instance, on the application of Figure 5a. By inserting a variable-ratio f quency divider between the VCO and the phi comparator it is possible to obtain various f quencies from the VCO using a single reference frequency f r Figure 5b. With this system a large number i frequencies may be obtained by a simpk method than in figure 5a, though at the ei pense of stability which generally decreases i Figure 6a. This VCO ci good linearity and wi up to 50 MHz. Figure 6b. This VCO circuit consists of a LC oscillator tuned and/or controlled by varicap diode. If the oscillator is also used ft tuning a receiver (i.e. as the local oscillator) is known as a tuneable voltage-controlle oscillator (TVCOI. Figure 8. The symmetrical multiplier is used ^ almost all PLL IC's and can also be obtain as an 1C in its own right. It may be co structed successfully from disc re components also. used, provided that the low-pass filter can vide sufficient suppression of the input quencies. This type of circuit is used in input section of an OTA and a PLL of performance can, in fact, be built with OTA type CA3080. Figure 10. If RF transformers are used, a chea multiplier may be built using four identic^ nrmnr pll systems elektor april 1975 — 415 Ire- Ire- 1 of tier I Illy for r) it lied the PLL. When it is to be used as an FM detector the linearity (Frequency change v. control voltage change) should be as good as possible, while for fre- quency synthesis this is unimportant but high stability is essential. Voltage-controlled multivibrators or varicap-tuned LC oscillators, like those shown in figures 6a and 6b respectively, generally have to be made up from discrete components, while integrated PLL circuits, such as the Signetics 565 shown in figure 7, rely on the triggering principle. Where a PLL is to be operated with a fluctuating supply voltage the VCO fre- quency should be independent of volt- age, or alternatively a stabilised supply may be used. b. Phase Comparator The output from the phase comparator or multiplier must be dependent solely on the product of the signals fed into it. This requirement is basically met by any non-linear component, subject to the proviso that the input signals also appear in the output. It is important to ensure that these signals have no detrimental effect on the performance of the system. An even more important requirement is that the output should not contain any D.C. components resulting from rectifi- cation of the input signals, as this can cause ‘mistracking’ and may even cause the PLL to go out of lock. If a balanced multiplier as shown in figure 8 is used impairments such as these can easily be avoided. The input signals are suppressed by the circuit and no rectification occurs. If suppression of the input signals is not required it is possible to use an asymmetric multiplier such as the example in figure 9. A circuit of this kind is included in the input of an operational transconductance ampli- fier (OTA) such as the CA 3080. This IC performs well in PLL circuits. It will be understood that rectification of the input signals can occur in this case, but nonetheless a satisfactory degree of AM suppression may be achieved. The best performance in this respect is achieved when the VCO output is fed into the asymmetric input and the input signal into the symmetrical input. The amplitudes of the signals should not exceed 0.5 V and 0.05 V respectively. The degree of AM suppression that may be obtained is almost as high as with a symmetrical multiplier. If R.F. transformers are available it is possible to use a diode ring modulator as a multiplier as in figure 10, but this is a rather old-fashioned method. The simplest, but unfortunately also the worst, solution for a phase comparator consists of a single semiconductor device that is fed with a VCO signal large enough to switch it on and off continuously. Because of the inevitable feedback from the circuit to the VCO a buffer stage is essential, as in the arrangement of fig- ure 1 1 . The phase comparator here is reduced to a mixer, so it appears that any mixer may be used as a phase com- parator. The problems that it introduces, however, cannot be eliminated without adjustment using expensive test equip- ment. Symmetrical phase comparators, on the other hand, give satisfactory re- sults with very little outlay on test equipment. c. The low-pass filter The low-pass filter (LPF) is the circuit that determines the bandwidth of a PLL. Simple RC filters, a few examples of which are given in figure 12, usually suffice. Examples b, c and d are suitable for symmetrical phase comparators, while a is applicable to asymmetric arrangements. As a general rule resistor R is already a component in the phase comparator. Although the calculation of component values for the low-pass filter is easily accomplished when using IC PLL’s by referring to the manufacturer’s data, sophisticated test equipment is needed to evaluate the performance of a PLL at frequencies in excess of 10 MHz. Filter d is the most suitable for home-built equipment. The cut-off frequency of the RC combi- nation formed by C 2 and the output resistance of the phase comparator is determined by the lowest frequency to be detected (20 Hz in Hi-Fi FM). The cut-off frequency of the second RC sec- tion, formed by P(at its maximum value) and C, both connected in parallel with the output resistance, is determined by the maximum PLL input frequency de- viation. Any desired bandwidth, up to a maximum determined by the loop gain and the input signal amplitude, may now be set with P. Problems experienced with PLL's Theoretically a PLL detector exhibits great advantages over other FM detec- tors, but in practice these are difficult to realise fully. There are two basic criti- cal factors: 1 . VCO frequency stability 2. Signal/noise ratio 416 - elektor april 1975 To obtain good stability the D.C. supply to the VCO must be temperature-com- pensated, and this applies also to the phase comparator if the control input to the VCO is asymmetric. In addition the components whose values affect VCO frequency should have zero tem- perature coefficients. These requirements are difficult to meet and in practice the VCO centre frequency often drifts sev- eral percent over the working tempera- ture range. For this reason it is advisable to choose the lowest possible working frequency. The lowest usable working frequency depends on the FM signal bandwidth and with the 200 kHz usual in FM broadcasting satisfactory opera- tion is possible with a working frequency as low as 450 kHz. Frequency drift at this low working frequency may be neglected; however, a receiver using this principle must employ double conver- sion techniques (i.e. it must be a double superhet receiver) and will inevitably cost more than a conventional receiver. Both the VCO and the phase com- parator generate some noise, so the de- modulated signal level must be as high as possible in relation to that noise. The PLL output-signal amplitude is pro- portional to the quotient of the devi- ation f and the working frequency, which in a receiver is of course the intermediate frequency fjp. With an intermediate fre- quency of 10.7 MHz and a deviation of 75 kHz this quotient is about 0.007, while with an IF of 450 kHz it is 0. 17 so that the lower frequency improves the signal-to-noise-ratio by about 28 dB. A PLL constructed from discrete com- ponents, working at 450 kHz and using the phase comparator of figure 8 and the VCO of figure 6a, can achieve a signal- to-noise-ratio of 60 dB on a stereophonic broadcast. Feedback PLL As outlined above, the main problem when using a conventional PLL as an FM detector arises from the standardisation on 10.7 MHz as an IF frequency. This means that practically all commercially available FM front-ends have an IF out- put at this frequency. In addition, special provision has to be made for the deri- vation of an automatic frequency cor- rection (AFC) control voltage from the PLL. However, by removing some of the components from the AFC loop in a conventional tuner the local oscillator can be used as a VCO. The linearity of such a VCO can be quite good since the 75 kHz deviation is small in relation to the working frequency (around 100 MHz). The reference frequency for the phase comparator can be supplied by a stable oscillator in which the fre- quency-determining element is a quartz crystal or a ceramic filter, so that VCO phase jitter noise, which is relatively strong at 10.7 MHz, is avoided. Figure 1 3 is a block diagram of a feed- back PLL. The aerial signal is mixed with the output from the tuneable voltage- controlled oscillator (TVCO) to give a 10.7 MHz signal that is fed through Figure 11. This circuit may be used as a phase comparator, but unwanted demodulation prod- ucts arise due to modulation. This precludes its use as an FM detector. Figure 12. Of these low-pass filter circuits version d is best for home construction as it is least critical to set up. Figure 11 Feedback PLL differs essentially from conventional PLL insofar as it includes the I F filter in the control loop. This results in a substantial reduction in the IF signal devi- ation, to the extent that the I F bandwidth can be low enough for m to be unity or less. This makes alignment of the bandpass filter and component values in the lowpass filter ex- ceedingly critical and for these reasons it is better to choose a larger bandwidth. There are a number of feedback PLL systems in which the principal aim is to maintain the modulation index as consistently as possible at unity. The complexity of such systems, however, as well as the difficulty of aligning them, limits their use to radio amateurs with sufficient theoreti- cal knowledge and to space-travel communi- cation. 13 pil systems decimal to bed converter elektor april 1975 — 417 FI Hits lie loll . an IF filter to the phase comparator. The other input to the phase comparator receives a high-stability 10.7 MHz refer- ence signal from the reference oscillator, thus, when the signal is locked in, the TVCO follows the aerial signal deviation. This means that the deviation of the 10.7 MHz signal is considerably reduced, hence the name ‘Feedback PLL’. Because of this reduced deviation the IF band- width is much smaller than in a con- ventional receiver. In the article entitled ‘Modulation Sys- tems’ the minimum bandwidth of an FM signal is given as: bmin. = 2(m+ l)f LFmax and this relationship is valid when m > 1. In a feedback PLL, however, the IF-signal modulation index is consider- ably less than 1 which accounts for the reduced bandwidth. The significant advantage of a feedback PLL system lies in the IF bandwidth, which becomes independent of deviation and in fact depends only on the highest modulation frequency. This gives improved signal-to- noise ratio and lower distortion com- pared to a conventional receiver, al- though the degree of improvement de- pends on the original modulation index of the aerial signal. For mono FM transmissions, with a maximum modulation frequency of 15 kHz and a modulation index of 5, the IF bandwidth in a conventional receiver must be 1 80 kHz, whilst the bandwidth in a feedback PLL receiver is only 30 kHz. The ratio is considerably less unfavourable for stereo transmissions however, as the highest modulation fre- quency of 53 kHz means that the feed- back PLL IF must have a bandwidth of 106 kHz. The principle of feedback PLL was known before the introduction of stereo FM broadcasting but unfortunate- ly this did nothing to prevent the intro- duction of multiplex stereo systems and so any improvements that might have been made in stereo reception were thrown away. It is still true to say, however, that a feedback PLL receiver similar to fig- ure 1 3 gives a considerable saving in cost compared to a conventional receiver with comparable performance. Feed- back PLL systems are of particular in- terest to radio amateurs, because signifi- cant improvements in signal-to-noise ratio may be realised if a low maximum modulation frequency is specified. How- ever, as far as the author is aware, little work has been carried out in this field. This is surprising as the principles in- volved have been known for many years and the VHF and UHF amateur bands offer unlimited possibilities for experi- mentation. Summary PLL is particularly suitable for frequency synthesis and for demodulation of FM signals. When used as an FM detector the relative deviation of the input signal should be as high as possible. This involves the use of multiple frequency conversion which is too expensive for the consumer market and too compli- cated for many home constructors. Feedback PLL’s may be used at high frequencies and offer the advantages of reduced IF bandwidth and lower dis- tortion with the absence of conventional AFC. Full exploitation of the potential of feedback PLL’s is probably too ex- pensive for consumer applications. Never- theless, simplified feedback PLL circuits are feasible and are indeed cheaper than conventional receivers. They should, therefore, be of interest in consumer electronics. VHF and UHF radio amateurs are par- ticularly well placed to take advantage of feedback PLL techniques, as their own experience makes them familiar with the RF work involved. A simple feedback PLL FM receiver will be described in a future issue of Elektor. M J. Wittje decimal to bed converter This converter can be used as a manual encoder which will convert decimal coded signals into BCD codes and drive digital circuits. Furthermore, the con- verter can be used as a teaching aid for explaining the BCD code. One 1C and six germanium diodes are sufficient for converting a decimal num- ber into a BCD number. A switch for zero is not provided because the converter automatically indicates zero when all switches are open. The reverse resistance of the diodes must be as high as possible (if necessary, check with an ohmmeter) and the gate inputs can be provided with a pull-up resistor connected to the posi- tive supply voltage. If the circuit is to be used to explain the BCD code, the BCD-output conditions can be indicated by means of LED’s. The circuit for the required buffer stage is shown in figure 2. H 418 - elektor april 1975 A. Seitz fido Fido is a new electronic game in which an unfortunate dog is called by four masters at the same time. The command "Fido come" is given by means of a pushbutton. At each push on one of the four buttons controlled by each player Fido jumps in the required direction. However, the four masters and/or mistresses have one handicap: After one successful command to Fido, the would-be Fido owner who has given the order has nothing more to say for a certain time. Then the other players can go on with Fido. If one of the players succeeds in getting Fido into his kennel, the game is decided: Fido stays where he is. Construction and operation l! Since Fido is clever enough to let him- I self be represented by a small incan- | descent lamp, he is not going to suffer I from an otherwise unavoidable nervous I breakdown. The worst that can happen is I that after a prolonged fight for mastery I over Fido our doggy will suffer from a I flat battery. I On the playing board nine lamps are I arranged in a square (figure 1). On the I extension of each side there is a lamp I representing a kennel (so in total four). | Furthermore, at each of the corners | there are four push buttons with a pilot lamp to indicate when a player can join | the game. The buttons make Fido jump i in four directions (away, towards, left or I right with respect to the particular | player). The photograph also shows that the “gaming table” is provided with an I on/off switch, an interval switch (coarse) and an interval control (fine) for setting I the obligatory rest period for the players, j These switches can be calibrated “blood- i hound/whippet” and “dog-tired . . . alert” I respectively. i Furthermore there is a switch to disable the “rest” lamps and there is also a starting switch. By pushing this button, I Fido takes up his position in the centre | of the field ; i.e. the middle lamp is alight. By pushing one of his buttons, each player can now try to direct Fido into his kennel. Once a player has pushed a button, he is obliged to take a breather before he can push a button again. The ! lamps fitted near the buttons indicate when the next command can be given. Each player can give only one command at a time. If an impatient player pushes his button too soon, the penalty is a new start of the waiting period. So Fido will not respond to a command that comes too early. To make the game a bit more exciting, the pilot lamps can be switched off, so that each player must just guess when he may next give a command. by nine lamps arranged in a square. These lamps are located at the intersections of 3x3 matrix rails. The signals for these rails are driven by two left/right shift registers. The clock pulses to the registers are produced by the players pushing one of the buttons. Since each player has four buttons at his disposal, Fido can be sent in all directions including the kennel of another player. The directing signals for left, right, up and down are coupled into the registers via the multiplexer. Once in a comer, the dog can be made to jump into the kennel situated below the comers as seen from the player’s position. The register input driving the “kennel” flipflop is so con- nected that the command for jumping is only followed if the other register, too, is in the proper position. The lamp field is blocked to prevent lamps from lighting up after a jump into the kennel. At the same time all register outputs are blocked so that no more “kennel” flipflops can be driven. The game is started by pushing the j starting button; then all the “kennel” I flipflops are reset and the two shift I registers take up a central position. In I that case the middle lamp is alight. The left/right shift register Figure 3 shows how a flipflop can be! turned into a “flipflopflap”. The inputs! of each nand are connected to the out-l puts of the other nands. Consequently,* only one output at a time can be low! (“0”). This “0”-signal produces a high! output level (“1”) at all the other nands;! these high levels in turn cause the low! output level on the first nand. A negative-* going pulse on one of the coupling rails! causes all nands connected to this rail tol change to “1”, whereas the nand whose! output is connected to this rail ensures® that this rail remains “0”. If gates with a so-called totem-pole® output are used (7400, 7420 and 7430W the outputs must be separated by means! a diode as otherwise none of the* 3 The block diagram Fido’s position in the field is indicated Figure 1. Artist's impression of Fido. Figure 2. The block diagram. The command- units also comprise the waiting time indication. The push button "start" resets all "kennel" flipflops and sets the registers at the central positions, so that the lamp in the center of the field lights up. Multiple connections between the circuits are indicated by means of broad Figure 3. The development of a multiple flipflop starting from the fundamental prin- ciple. a. Two methods of drawing a simple flipflop b. A 3-fold flipflop c. A 5-fold flipflop. „ j outputs would change to low (figure 3c). Ll” With types with an open collector output l? t I this is not strictly necessary, although n I it is recommended to keep the input load I of the pulse low. I In that case the “0” must, after each J pulse, shift one position to the right, I left, top or bottom. So we need a memory which remembers what coupling rail is carrying a “0” signal before the pulse, and a circuit that determines in what direction the shift should take place. The memory is formed by Ci, (C 2 , C 3 , figure 4); the direction of shift is deter- mined by two nands(N 3 and N 4 , N 5 and N 6 , N 7 and N g ) which receive their sig- nals via N] and N 2 . When the button is pushed, say left, this is what happens: Via Nj, connected as an inverter, the “ 1 ” signal is fed to the nands N 3 , N s and N 7 via the “left” conductor. At the same time all the connecting rails are brought to the “0” level via the diodes Di , D 3 , D 4 and D 5 . As a result, the nand N 9 , 420 — elektor april 1975 fidt Figure 4. Complete "shift register for a zero", 3-fold, for the matrix line of the horizontal shift register. The vertical register is of the same construction (description between brackets). Figure 5. Field with waiting time indication. Depending on the type of field used, the trigger unit is required several times. It serves to sup- press contact bounce. Figure 6. Diagram for Fido with nine lamps. If the whole is fed from batteries, it is advisable to supply the lamps from a separate battery because pulses caused by switching (low filament resistance of an extinguished lamp) might interfere with the circuit. The bias of Cg (figure 5) must also be obtained from a separate battery because a maximum current of about 200 mA can occur. Nio or Nn , which has been at “0” level so far, changes to “1”. Simultaneously, a positive pulse is fed to the two adjacent nands via the capacitor connected to this output. The gate thus prepared by the “1” signal via the conductor “left” maintains the collecting line of its neigh- bour at “0” until again via diode D] the “0” signal disappears and the remaining conductors become logically “1”. The contact potentials of the diodes Di and D 3 up to and including D s ensure that the coupling rails reach the “1” potential before the inputs of the gates 1 or 2. This is necessary to ensure that the new main nand takes over the “0” signal before the direction determining gate changes back to “1”. In the extreme positions for the “shift register for a zero”, the “kennel”- flipflops Ni2-N, 3 and N, 4 -Ni S are driven. These may be driven only when the sec- ond register reports the correct position. The outer direction determining gates Nj and N g , which drive the “kennel” flip- flops require three inputs for that pur- pose; one being coupled to the corre- sponding matrix line of the other register. Command-unit with indication Figure 5 shows a command-unit with four push buttons. The other units are similar. Via Pi and R 17 or Rig, respectively, capacitor C 6 is negatively charged until the voltage across C$ equals the sum of the contact potentials of diode Di 0 and the base-emitter junction of T 9 . The latter is then conductive, so that Ti 0 causes the lamp to light up. The pilot lamp indicates when a command can be given. The waiting time can be adjusted with Pi . When pushing a button, say Si, Tj is turned on by the negatively charged capacitor C 6 , so that the emitter of T i drops from +4,5 V to +0,7 V. This pulse serves to drive the shift register. Due to contact bounce, Fido is likely to make wild and unpredictable jumps, or just stays where he is. To avoid such “disobedience”, each push button must be connected to a trigger. Even the shortest pulse at the base of Ti is suffi- 422 — elektor april 1975 fido cient to cause the two transistors (T L and T 2 ) to switch. As a result capacitor C 6 is connected to the control line until the voltage drop across Ri 3 caused by the charge current is no longer high enough, and the trigger returns to its initial position. Then capacitor C 6 discharges across R )7 (Rig) and P t . The complete diagram Owing to the large extent of the circuit, some of the sections are represented as blocks in figure 6. The positions indicated by the coupling rails are represented by “0”-signals. For the remainder, only “1 ’’-signals are used; hence the inverters 7405 for inverting the signals. These signals are fed to the lamp drivers 7440 which cause the lamps to light up when all inputs are “1”. Since only two of the four inputs of the lamp drivers are used, all the others can be connected to the positive of the supply, which, however, is not necessary. Once Fido has disappeared into a kennel, that is to say. when a “0” signal has reached the input of a goal flipflop, a “ 1 ” is produced at the driver of the goal lamp, and a “0” at the gate N 20 , which via the inverter N 2 | and six diodes Du up to and including D !6 trans- fers this signal to the outputs of the inverters Ii up to and including I 6 . As a result all the lamps in the field are ex- tinguished. Furthermore, all the outer direction-determining gates (figure 4) are blocked (“0”-signal at the inputs that are connected with the inverter outputs), so that no further goal can be scored by the now invisible Fido, if more but- tons were pushed. The start- or reset button returns the goal flipflops and the registers to their initial positions again. The middle coupling rails must be connected to the reset conductor via the diodes (D 9 in figure 4) . The words “left”, “right”, “top”, “bot- tom”, “vertical” and “horizontal” are related to a group of push buttons which is fixed by an arbitrary position of a player and is called command-unit 1 . The other command-units are numbered clock-wise. The arrows in figure 5 are related to the way in which Fido moves as regards the player concerned. Variations The game can easily be changed. A first possibility is to expand the field so that the game will last longer (figure 7, accord- ing to the principle in figure 3c). This will, of course, increase the cost of the unit by a considerable amount, especially if the 25 lamp version of figure 7 is used. Furthermore, it should be noted that the field is in fact only suitable for four or eight players, whereas the smaller field can also be used by two without Fido endlessly running up and down. On the other hand, the field with 25 lamps can easily be connected to eight command-units, so that eight “dog lovers” can join the game. A “mini Fido” is also a possibility if we restrict ourselves to one register (see fig- ure 3c), and if the “kennels” are placed at the two ends of the row of lamps (fig- ure 8). In spite of the simple set-up the game can still be fun; playing with the push buttons alone is most amusing. In addition this version offers the possibility of studying the register. Of course, other possibilities can be worked out, but then again it is up to the reader to find an arrangement in accord- ance with his taste and, lets face it, budget. M elektor services to readers EPS print service Many elektor circuits are accompanie by printed circuit designs. Some of thes designs - but not all! - are also availabl as ready-etched and predrilled boards,! which can be ordered from our Canter- bury office. A complete list of the) available boards is published under the heading ‘EPS print service’ in every issu< Delivery time is approximately three weeks. As a further service, boards which are taken off the regular service list at somi future date will continue to be avaiHblt in spite of this: delivery time will thei be approximately six weeks. It shoult be noted, however, that only board which have at some time been publish in the EPS list are available; the fact tha a design for a board is published in particular article does not necess imply that it can be supplied by elekto Technical queries Members of the technical staff will available to answer technical queri (relating to articles published in elektor by telephone on Mondays from 14.00 16.30. Queries will not normally answered at other times. Letters should be addressed to th; department concerned: TQ = Technic Queries. Although we feel that this is a essential service to readers, we regret tha certain restrictions are necessary: 1. Questions that are not related t articles published in elektor cann- be answered. 2. Questions concerning the connectio of elektor designs to other units (e. existing equipment) cannot normall" be answered, owing to a lack of pract' cal experience with those other unit' An answer can only be based on comparison of our design specifi cations with those of the other equip- ment. 3. Hieroglyphs or illegible handwritin cannot be decoded: provided the sender’s address is legible, the lette is returned unanswered. 4. Questions about suppliers for com- ponents are usually answered on the basis of advertisements, and readei can usually check these themselves. 5. As far as possible, answers will be on standard reply forms. We trust that our readers will understand! the reasons for these restrictions. On the one hand we feel that all technical queries should be answered as quickly® and completely as possible; on the othei hand this must not lead to overloading] of our technical staff as this could lead to blown fuses and reduced quality future issues ... elektor april 1975 : The block diagram of the interval switch is given in figure 1 ; it consists of a pulse generator, two monostable multivibra- tors and a stabilizing circuit. A mech- anism controls the automatic dia- phragm and shutter of the camera. The pulse generator consists of a UJT (unijunction transistor) relaxation oscil- lator with adjustable pulse recurrence frequency. The output pulse drives two interconnected monostable multivibra- tors (MMVs) which control the mech- anism for diaphragm adjustment and H.U. Heinz A A machine 16 Slowly-developing technological processes or natural events cannot be perceived because the eye is generally not able to distinguish the separate stages. Such events and processes can, however, be visualised by means of cinematographic time compression. An interval switch linked with a camera enables it to make single exposures at set intervals. When run at normal speed the film then shows a process or event apparently developing continuously, but in a much shorter time. camera shutter. Because the circuit must be suitable for battery supply, a stabil- izing circuit ensures a constant voltage throughout the battery life. Of course, the circuit can also be fed from a mains power supply. MM V 1 controls the automatic diaphragm of the camera. This diaphragm setting is maintained until MMV2 operates the shutter and resets the entire circuit to its initial state. The stabilizing section included a battery voltage indicator which operates Photograph 1. The time compressor system for film cameras. The box mounted on the camera contains the relays and the shutter drive motor; the box beside it contains the electronics. with an ‘expanded scale’ and ‘suppressed zero’ so that it only reads from about 12-20 V. Since the circuit will not function correctly if the battery voltage falls below 12 V, there is no point in measuring below 12 V. It is simply a waste of meter scale space. The pulse generator Figure 2a shows the principle of the pulse generator. Capacitor Ci charges via Ri to the breakdown voltage of the UJT, to discharge again via resistor R 2 424 — elektor april 1975 I and the E-Bi junction of the UJT. The I breakdown voltage of a UJT is an almost fixed percentage of the supply voltage; | usually between 60% and 85%, depend- . ing on the type. I * Positive pulses appear across resistor R2 , with a repetition frequency that can be ■ adjusted within certain limits by changing R], In the circuit of figure 2b, Pj is the potentiometer with which the repetition frequency is adjusted. The adjustment range of Pj is determined by the series connection of Ri and P2 in parallel with Pj. Via the selector switch S2 this combination is connected to the series circuits R 3 + P 3 . . . R 6 + P 6 which are connected to the supply. I Terminal B2 of the UJT is connected to the supply via resistor R 7 . This resistor serves to reduce the temperature dependence of the UJT. I In the blocked condition, the E-Bi junction of the UJT has a very high resistance so that it is possible to achieve relatively long pulse times with large capacitances (220//) and high resistances (maximum 1 M). Switch S, j is combined with the on/off switch; in the centre position C2 charges rapidly via R 3 , so that the UJT can I produce the first pulse the moment the I on/off switch is operated. If the capaci- | tor were not given an initial charge in this way, the waiting time for the first pulse would be 4 minutes in the worst case. Transistor T2 serves as an inverter, so that the pulse generator supplies both positive and negative pulses. The Monostable Multivibrators (MMVs) j The two MMVs connected after the I pulse generator are equipped with 1 thyristors with anode- and cathode-gates I because these can fire on positive as well as on negative pulses. Both MMVs are of the same design, differing only , in component values. I Figure 3 shows the circuit of an MMV. I Once thyristor Thj has been fired by negative-going pulses on the anode-gate, it remains on until the current drops below the so-called holding current. If in the anode circuit of the thyristor a resistor is included of such a value that the holding current of the thyristor cannot be reached, the thyristor will not fire. If, however, a capacitor (C4) is now | connected parallel to this resistor, the thyristor will fire and the capacitor will begin to charge. Since, however, the charging current of a capacitor decreases as the charge increases, there comes a certain moment when the current flowing through the parallel circuit of resistor and capacitor drops below the holding current, and the thyristor blocks again. The capacitor then discharges through the parallel resistor R 10 (figure 3). A variable series resistance (P 7 + R n ) determines the charging time of the capacitor and thus the time during which the thyristor remains on. In addition, this series resistance protects the thyristor against excessive switch-on currents. Via R 14 and D, the thyristor drives switching transistor T 3 which energises relay RLA. Diode D2 protects the transistor against voltage surges when the relay cuts out. Current supply and measuring circuit The supply voltage is stabilized at about 1 1 V by ZDj and T s (figure 4). All battery voltages can be measured under loaded and no-load conditions via switch S 4 . As long as the measured voltage is higher than the zener voltage, a cur- rent I flows through the parallel circuit (R22 + P12); the resulting voltage drop is measured with the measuring instru- ment. The meter is adjusted to full-scale deflection (f.s.d.) by means of P 12 . The currents through the zener diodes ZD2 . . . ZD 4 can be adjusted with the potentiometers P 9 ... Pa - These zener diodes ensure that only voltages higher than the minimum voltages on which the apparatus functions properly are measured. The meter thus has a ‘suppressed zero’, i.e. it only reads from (say) 12 V upwards since voltages below this are of no interest. The whole meter scale may then be calibrated for 1 2-20 V. The residual battery charge can be esti- mated on the basis of the difference in meter deflections when readings are taken with and without load. The extra positions on S4 are for testing other batteries in the camera. The diodes ZD 3 and ZD4 can be chosen to give a suitable ‘suppressed zero’ value for other battery voltages. The complete circuit The complete circuit given in figure 5 is intended for a Zeiss G.S-8 synchron- ous camera. In this case the diaphragm is adjusted by a motor, so that it remains in the set position when the control current is switched off. The camera is fitted with two external connections for electrically-operated remote release ; one for single exposures and one for running exposures. Before the release is operated, the diaphragm must be properly adjusted. time machaj The negative pulse produced at the collector of T2 first starts MMV1 which, via RLA1 (figure 6) switches on the automatic aperture control for aboil 2 sec., giving ample time for this contra to find its setting before the shuttei opens. The moment MMV resets, i positive pulse starting MMV2 occurs a the anode-gate resistor (R13). As ; result RLB is activated, closes contact* RLB 1 , and starts a motor which drive the camera shutter. Although RLA is no longer energised the diaphragm motor will hold the aperture at its correct setting. The dia- phragm drive can be switched off altogether with S 3 , so that, for example; an electronic flash can be used with 1 preset aperture. S 5 operates RLB directly and can therefore be used foi manual shutter operation. There are almost as many automata exposure devices as there are camen types. Consequently the matching ol the automatic operating equipment to the camera diaphragm and shutte mechanisms often calls for considerabl Another type of automatic exposui control which is found in most camera nowadays uses a moving coil (as in meter) to control the diaphragm accord ing to the photocell response. In thi case, the circuit operating the diaphragu control must remain switched on whil the shutter opens. This can be achieve by providing an extra pair of contact Figure 1. Block diagram of the time com pressor. Figure 2a. Circuit diagram of a pulse genet ator using a UJT. Figure 2b. Diagram of the pulse generator. Figure 3. One of the two MMV’s with whie the diaphragm control and shutter ar operated. Figure 4. This stage serves for voltage stabil izing and checking the operating condition of the batteries. i i/nmin is also required to function for these shots, three components must be added to the cathode gate circuit of Th 2 : a 3n3 capacitor, a 470 12 resistor and a diode (DUS). This must be done in the same way as with MMV2 (here it is Cs, D3 and Ris). The push-button of figure 7 must then be connected direct to the additional capacitor. If the current consumed by the auto- matic exposure control is known to be small, the control can be left on con- tinuously during time-compressed filming. It will then be possible to dispense with T 2 and associated components, as well as with MM V 1 and RLA. One pair of contacts on RLB will suffice. It can be gathered from what has been said that adapting the circuit to a particular make and model of camera not only calls for a precise knowledge of the camera; it also requires consider- able experience in the field of elec- tronics. Anyone who undertakes this 1 project should be capable of tackling | Aligning the circuit Before the apparatus can be used, the following adjustments must be made. 1. Pi to zero, P3 to give maximum pulse interval. This will be about 2 sec. for a mechanical shutter and about 0.5 sec. for an electric shutter. 2. P4 , Ps , P6 to 1, 2, 3 minutes respect- ively. 3. Pi in position ‘maximum’. Adjust P 2 until the difference between the minimum and the maximum positions of Pi corresponds to 1 minute. 4. P 7 to a time which enables the automatic exposure control to re- adjust by two stops. 5. Pg to give the minimum time the shutter mechanism needs to operate the shutter when the battery is low. 6. Adjust Sj and S4 to ‘off position, P9, Pm and Pn to give 2.5 ... 5 mA measured between the contacts of S4 . Adjust Pi 2 to full-scale deflection of the meter. When choosing the zener dioi (ZD 2 . . . ZD4) take into account minimum voltages at which the equ ment will still function properly at 1 temperatures. If the zener voltages changed, other values may have to chosen for the adjustment potent meters. Figure 6. Relay contacts for cameras motor-driven or moving-coil diaphr control. nr iin time machine elektor april 1975 - 427 The shutter mechanism (for mechanical shutters) As is apparent from the previous examples, cameras with electric shutters are easily modified by bridging the the release contacts by the relay contacts. With mechanical shutters however the release button must be operated by a servo or other device. No detailed data can be given on the release mechanism because the con- struction depends largely on the camera used. The author used a Graupner Varioprop-Servo from which the feed- back potentiometer had been removed. This was used to drive the shutter release via a Bowden-cable type remote release. Limit switches were incorporated to limit the servo travel. A model control servo which may be adapted to a shutter drive for most cameras will be obtainable in a shop for model builders. Exposures with the time com- pressor To conclude with, some remarks about the exposure technique. To ensure a flowing motion, calculation of the intervals should be based on 900 frames, so that at a projection speed of 1 8 frames per second the projection time is 50 seconds. If the interval is indicated as t seconds per frame (F), and the time in which the compressed event takes place is T hours, we have: in which T is in hours, and t is in seconds. For an opening rose the interval for an exposure time from 0530 to 2030 (exactly 1 5 hours) is t = 4 x 1 5 = 60 seconds per frame. When filming outdoors, don’t forget to immobilize the flower in case it should sway in the wind. Editorial note A number of notes as regards component values may be made: All electrolytic capacitors must be of the 16 or 25 V type. For T 2 a BC 140 may be used instead of a BFY 39. Furthermore, it is advisable to connect a resistor of 1 k in series with the base of T 2 . In figure 4 transistor Ts (2N2219) may be replaced by a BD 137 or BD 139. In many cases this transistor will also have to be cooled, certainly if the two relays draw con- siderable current (over 100 mA). Finally it should be noted that in figure 4 '+V b ' is the output of the stabilized supply. So this point is the supply point ('©■) in figures 5 and 7. The voltage is about 11V. M marine diesel Apart from ship sirens and fog horns, builders of ship models are also interested in imitating marine engine noises. With only a few components the ‘marine diesel’ circuit lends realism to a model. The noise produced by a diesel-driven ship is made by the thump of the engine and the regular puffing of gases escaping through the exhaust. The noise of these escaping gases is imitated by a small noise generator in the circuit. The thump effect is achieved by using an 1C in a trapezium generator circuit, with the noise added on the leading and trailing edges. The figure shows the circuit. The base-emitter junction of Ti is reverse biased to break- down and the resulting noise signal is fed to the non-inverting input of the opera- tional amplifier. The feedback network, formed by R 4 , R s , R 6 and C 3 then determines the form of the trapezium voltage. As long as the IC has not reached saturation, the output produces a voltage ramp with superimposed noise. The noise is suppressed as soon as the IC reaches saturation. An oscilloscope connected to the output of the circuit should show one of the waveforms drawn in the dia- gram, depending on whether the DC- connected or the AC-connected oscillo- scope input is used. If after completion of the circuit it is found that the sound produced by the model is too slow, certain modifications may be made. Ci affects the noise; C 2 , Rs and R 7 determine the repetition rate. The output of the circuit can be connec- ted to the input of an amplifier. A resistor (value to be found by experiment, de- pending on amplifier sensitivity andinput impedance) connected between the cir- cuit and the amplifier prevents overdrive of the amplifier. 14 noise generator Despite its simple design, this circuit is a universal noise generator which pro- duces a very high noise amplitude. Transistor T! is connected as a zener diode and is connected to the base of the second transistor (T 2 ). The current through the zener transistor, and hence the amplitude of the noise, is adjusted by resistor R, . This noise voltage is then amplified by T 2 . The supply voltage can be varied over a wide range and, depending on the re- quired output voltage, can be chosen between 10 V and 30 V. At a number of different supply voltages the following noise output voltages were measured: +V b = 12 V — 5 Vp. p + V b = 1 5 V — 8 V m + V b = 20 V — 10 Vp. p + V b = 25 V — 15 Vp- p If required, transistor Ti serving as the zener diode can, of course, be replaced by a real zener of 6-8 V. port 2 minidrum The Minidrum described in the previous issue may, by the addition of various extra circuits, be extended to a comprehensive manual drum kit. Some new instruments, a three channel ruffle system and an automatic bassdrum are described in this article. The basic Minidrum contained only three instruments, a bassdrum, a snare- drum and a cymbal and so only three channels of the TAP were used. Since the TAP board has facilities for six channels the design example given here is based on six instruments. The number of instruments may, of course, be ex- tended to suit individual taste by adding extra TAP boards, one for every six additional instruments. A pulse generator is included in the design. This is intended to drive the automatic bassdrum, but may be used to drive other instruments either separ- ately or simultaneously. The ruffle system comprises three ruffle channels driven by a single oscillator. A pulse train appears at one of the outputs when a finger is placed on the appropriate touch contact. This may be used to drive any of the instruments to give drum rolls etc. The first part of the Minidrum to be described is the TAP circuit which controls the instruments via touch con- tacts. The Minidrum TAP Figure 1 is the circuit diagram of the complete Minidrum TAP. It has six touch inputs and six outputs, corre- sponding to the six instruments used in the design. As described in the previous issue each TAP channel consists of a COSMOS inverter (1 j-I«) followed by a diode and an integrating network. Hum from the player’s skin causes the output of the inverter to switch between logic 0 and 1 , charging capacitor (Cj-Cg). This output voltage controls the relevant instrument. The 47 k resistors (R 7 -Ri 2 ) limit the base current of the one-shot associated with each instrument. Two types of RCA COSMOS IC may be used for the TAP, CD4009AE or CD4049AE. When using the former diode Di must be included in the circuit (see figure 1), if, however, the CD4049AE is used, Di may be replaced by a wire link on the p.c. board. Due to the high noise immunity and wide supply voltage tolerance of COS- MOS circuits a sophisticated power If if If 'MINIDRUM TAP' SSS 430 — elektor april 1975 supply is not required, although some form of simple stabilizer is desirable. The Minidrum TAP p.c. board Figure 2 is the p.c. board and component layout for the TAP circuit of figure 1. It is recommended that a socket be used for the IC to avoid the possibility of damage due to static or leakage from unearthed soldering irons. A photograph of the completed board is given in figure 3. The instruments All the percussion instruments use the gyrator board described in last month’s issue, the circuit of which is given in figure 4, but with component values to suit the different types of instrument. Table 2 gives the component values which are common to all the gyrator boards, while table 3 gives the compon- ents which determine the characteristics of the individual instruments. The gyrator p.c. board and component lay- outs are given in figures 5-13. Each percussion instrument has two inputs, input 1 and input 2. A mono- stable multivibrator (one-shot) is con- nected to each of these inputs and these one-shots drive the gyrator. The output from the TAP is used to drive input 1 while input 2 may be driven by the ruffle system if desired. If ruffle is not required on a particular instrument then the monostable on input 2 may be omitted, as in last month’s article. As described in last month’s article, the snaredrum has filtered noise mixed in with the output of the gyrator. The cymbal, brushes and maraccas are merely filtered noise, with no gyrator input. The p.c. board given in the previous issue is used for the noise generator and noise gating. If an instrument is to be used with the ruffle system (the snare- drum for example) then both gating inputs of the snaredrum noise board are used, one for the manual input and one for the ruffle input. In the case of the snaredrum these inputs are driven by the one-shots on the gyrator board and thus the one-shots on the noise board may be omitted (figure 15). In the case of purely noise instruments (cymbal, brushes and maraccas) the manual TAP drives the noise board directly and the one-shot(s) must be used (figure 14). If the ruffle system is not used then one noise board will do for two instruments, as was the case in last month’s article where snaredrum and cymbal noise were produced by the same board. The board and component layouts are given in figures 16-20. The component values for the maraccas and cymbal noise boards are given in tables 4 and 5, those of the snaredrum noise board in table 6. R x is added in the circuit for the brushes. To mount this resistor on the p.c.b., the connection Rm - C 2S is left ‘in the air’, and R x is connected between this junction and the original connection to T is (see figure 19). The automatic bassdrum Figure 21 is the circuit of the pulse Figure 4. Circuit of the complete gyrator board with two input monostables. The component values are listed in table 3. Figure 5. The gyrator p.c. board. Figures 6-13. Component layouts for all the gyrator instruments listed in table 3. generator for the automatic bassdrum. The circuit is desgined around an RCA COSMOS IC, the CD4011AE, which is a quadruple two-input NAND gate. Gates Ni and N2 form an astable multi- vibrator. Pulses from the output of Ni are inverted and squared by Tj. C4 and Rs differentiate these pulses and D2 clamps the output so that only positive going pulses appear at the cathode of D 3 . These pulses may be used to trigger any of the instruments, but in the system described here they are used to drive only the bassdrum. The tempo of the bassdrum may be adjusted from about 40 to 240 beats per minute by means of Pj. In passing it may be noted that the circuit of figure 4 may be used on its own as a metronome, by reducing R3 to 15 k, R4 to 1 k and Rs to 4k7. C4 is increased to 1 00 n and D3 is replaced by a 47 S2 resistor. The circuit will then drive a small loudspeaker directly, and may be used as a self-contained unit with a battery since power consump- tion is quite low. Instead of a mechanical start/stop switch the automatic bassdrum of course uses a TAP. N 3 and N4 are connected as a set-reset flip-flop; touching the start contact sets the flip-flop and touching the stop contact resets it. In the reset (stop) condition the output of N 3 holds the inputs of N 2 high via Di and the astable will not start. In the set (start) condition the output of N 3 is low and Di is reverse biased, so the astable runs. The circuit is so designed that as soon as the button is touched the circuit produces its first output pulse, even at low repetition rates. When the stop button is touched the circuit stops immediately. Figures 22 and 23 give the p.c. board and component layout for the bassdrum pulse generator. Again it is recommended that a socket be used for the IC. Figure 24 shows a photograph of the completed board. The ruffle system The circuit of the complete ruffle 432 - elektof april 1975 Figure 14. Noise circuitry for the Cymbal, Brushes and Maraccas. See table 5 for the values of the unmarked components. Figure 15. Snaredrum noise circuit. Note the absence of input monostables. Table 4. Components list for Cymbal, Maraccas and Brushes, (figures 14 and 17-19) for components common to all boards. Where values differ see table 5. Resistors: R 42. R 53. R 58. R 60- R 70. R 75 = 10 k R44. R45 . r 48- r 63- r 65- r 73 ■ 470 k R 49- R 66 = 6k8 R 50' R 67 = 888 k R 52- R 69 = 5k6 R 55 , R 72 = 27 k R 56- R 76 = 100 k R74.R77 = 270 k R59 = 4k7 P 2 = 10 k preset Capacitors: C-)g = 100/1/10 V Semiconductors: Tl2' T 13- T 16- T 17.Tl8. T21. T 22.T23 = T UN Tl4.Tl5.Ti9 = TUP D7. D 9.°10.Dl2. d 13- d 14 = DUS Table 5. Components list for Cymbal, Maraccas and Brushes, for components unique to one particular instrument. Cymbal S i Brushes R 43' R 61 10 k 10k 2k 7 27 k 10k 180 k 820 k 220 k 820 k 10 M 10 M - 220 k 270 k 220 k r 54 100 k 8k2 10 k 470 k 470 k 270 k 100k 8k 2 100k R x 180 k 150 n 100 n 100 n Cl8 68 n 120 n 39 n C19.C26 12 n lOOp 47 n C20.C27 100 n 12 n X 4n7 680 p 680 p 68 n 470 p 390 p C23 4n7 470 p 390 p C 2 4 150 n 100 n 1 A* 68 n 120 n 180 n 4n7 680 p 100 p C29.C3O 100 p 680 p 100 p 10 n 10 n 2n7 D8.D11 DUS DUS — = wire ink Table 6. Components list for the snaredrum noise board (figures 15 and 20). Resistors: r 78- r 91 = 820 k R 79- R 92. R 99 = 478 k r 80. r 93 = 6k8 r 81> r 94 = 880 k R 81a. R 94a= 18 M R 82- R 95 = 188 k r 83- r 96 = 5k6 r 84. r 88 r 97. r 101 = 18 k r 85. r 98 = 15 k R 86 = 4M7 r 87. r 102 = 188 k R 89 = 4k7 Rl00. R 103 _278k P3 = 10 k, preset Capacitors: C33 = 100 H. 10 V c 34- c 39 = ®n2 c 35- c 40 = 22 n C 36- C 37.C41.C42 = 2n7 C3S- 1n2 C43.C44 = 10 n Semiconductors: Di5,Di5,Di7,Di 8 ,Di 9 , D20'^21>^22 = DUS t 24- t 25. t 27- t 28 = T UP T2e.T29.T30.T31 = tun ' 1 ■I 1 1 1 1 1 f llif m T 1 1 r r r r elektor april 1975 Capacitors: Ci = 100 H, 10 C 2 = 27 n C 3 = 2/J2, 10 V C 4 = 2n7 Semiconductors: 1C = CD401 1AE Ti = TUN D, = BAY 61, BA 220 D 2 ,D 3 = DUS Figure 16. Noise p.c. board. Figures 17-19. Component layouts for Cym- bal, Maraccas and Brushes respectively. Figure 20. Component layout for snaredrum IC1=N1...N4=4011 Figures 22 and 23. The board and component layout for the bassdrum pulse generator. system is given in figure 25. The system is very similar in operation to the automatic bassdrum. i! and 1 2 form an astable multivibrator and 1 3 serves to buffer the output and improve the waveshape of the astable. I 4 -I 6 form the TAP control for the ruffle system. As the three channels are identical only one will be described. When the touch contact is not being touched the input of I 4 is held low via The output is therefore high. C 4 is charged via R 3 and current flows into the base oPT 2 via D 2 and R 12 . T 2 is driven into saturation so that the ruffle signal appearing via D 6 is blocked. When the contact is touched the output of I 4 switches at 50 Hz between ‘0’ and ‘1’ BD PULSE GEN. 436 — elektor april 1975 elektor april 1975 - 437 Components list for figures 25 and 27. Rl.R2. R 4. R 6- R 8 = 10 M R3.R5.R7 = 47 k Rg,RlO. R 12< R 13- R 15. R 16 = 10 k Capacitors: Ci = 220 /Li/1 0 V C2.C3.C7.C8.C9 = 4n7 C4.C5.C6 = 1 *t 5/10 V (2*12/10 V Semiconductors: 1C = CD 4009 AE or CD 4049 AE Tf . . . T 3 = TUN D-, . . . D 15 = DUS Z x = Z-Diode (see text) Input resistors for mixer preamp (figures 29 and 30). Figures 26 and 27. The p.c. board and component layout for the ruffle system. Figure 28. The completed ruffle board. Figure 29. Circuit of the mixer-preamplifier. boards given in this example is given at the side of the figure. It should be stressed again that this is only an example and that any combi- nation of instruments may be used to suit personal taste. All that is required is a little common sense and application of a few simple rules. When choosing the combination of instruments the following points should be noted. 1. For each gyrator instrument one gyrator board is required. 2. For the noise instruments (maraccas, brushes and cymbal) one board will do for two instruments, unless ruffle is required, in which case one board is required per instrument. If ruffle is not used then the input monostables are included on the board and the instru- ments are driven direct from the TAP. If ruffle is used then the monostable on the input driven from the ruffle board is omitted. 3. In the case of the snaredrum, if ruffle is used then one noise board is required for this instrument, both input monostables being omitted and the inputs driven from the ruffle board Figure 31. Example of a complete drue system with four gyrator instruments and twe noise instruments. Figures 32 and 33. Photographs of a comprt hensive manual drumkit using all instrument except brushes. 1 I I minidrum elektor april 1975 - 439 than clear Perspex as this will afford some electrical screening. In figure 32 three noise boards for the cymbal, maraccas and snaredrum are on the left. To the right of them are seven gyrator boards and on the extreme right the auto bassdrum. Along the bottom of the photograph are the two TAP boards and the ruffle board. The mixer- preamplifier described in last month’s article is at the top right-hand comer. The circuit and board layout are given in figures 29 and 30. The mains transformer should be mounted well away from the TAP and ruffle boards and the mixer-preamplifier to avoid hum pick-up. Note that 9 chan- nels of the TAP or I'A boards are used in this example. The Minidrum will be on display (and work- ing) together with many other Elektor pro- jects at the 1975 London Electronic Com- ponents Show at Olympia, May 13-16. 440 — elektor april 1975 compressor £A Compressors are now being used on an ever-increasing scale. They may be found in tape recorders, intercom systems and baby alarms, public address systems, disco- theques and of course broadcast transmitters. A compressor supplements a manual volume control and allows a system to adjust itself to a wide range of input signals with little distortion. The design described here should find a wide range of applications with the electronics enthusiast. The aim of compression Where signals with a wide dynamic range | have to be processed it is desirable that ] as little distortion as possible should LI occur. The designer of, say, a public |j address system may have given much I thought to achieving a good distortion *1 figure, but this is of no avail if the I system is overloaded by an enthusiastic speaker shouting into the microphone. j It is of course possible to prevent a II circuit from being overloaded by li attenuating the input signal with a |l fixed or manually variable attenuator, |l but then in the example above the I person who mumbles into his notes | would certainly not be heard. | This is where a dynamic range com- il pressor comes in. A compressor is I basically an attenuator, or variable gain II amplifier, which is controlled by the If signal it is attenuating, either directly || or by a control voltage derived from the |l signal. As the signal increases so does I the degree of attenuation, so the com- I pressor tries to keep the output signal constant whatever the input. This can- not be achieved in practice, but it is possible to limit the output to a narrow range over a wide range of input signals. In a p.a. system (figure 1) a compressor could be included between the micro- phone preamp and the normal volume control. The compressor, like death, is a great leveller. Compressor Transfer Functions At first sight it would seem to be an admirable aim to control the output signal amplitude with the input signal as in figure 2. This system has an overall gain of - where K is a constant and vj is the input voltage (for an attenuator of course the gain is less than 1 ). So V 0 = — = K. vi The output voltage is therefore constant for all input voltages. This seems admi- rable until one considers what happens 1 olume 'i — — lyf v c Figure 2. A first approach to a transfer function for a compressor. This is doomed to failure however. Figure 4. a. Voltage-current curve of a fila- ment lamp. The resistance increases with increased current. b. Compressor using a lamp and a fixed resistor. c. T ransfer function of the compressor. Figure 5. a. Voltage-current curve of a VDR. b. Compressor using a VDR and a fixed resistor, c. Transfer function of the com Figure 6. Dynamic characteristics of variou types of compressor in response to a sudden burst of signal. Figure 7. Block diagram of an active com pressor using a peak detector to derive I control voltage which alters the attenuator. elektor april 1975 — 441 when vj is zero. The gain then becomes infinite and this idea becomes unnat- tractive. A much better solution is to control the output signal with the output signal, which at first sight may seem odd. In figure 3 however it can be seen that the gain is — . v 0 , Kvj Therefore vo = — . 2 v ° or vo = Kvj This is a square-law compressor func- tion. Of course, other functions may be achieved, notably logarithmic, where vo = K log vj. Practical Compressor Circuits There arc many different kinds of com- pressor circuit. One of the oldest and simplest circuits makes use of the non- linear resistance of an incandescent lamp, whose resistance increases as the current through the filament increases. In figure 4 the resistance of the lamp, which forms the upper limb of the attenuator, is low at low signal levels so only a small portion of the signal voltage is dropped across it. At higher signal levels the resistance increases and a larger proportion of the signal voltage is dropped across the lamp. The output signal therefore does not increase as much as it would with a normal attenu- ator. The thermal inertia of the lamp filament means that this circuit cannot follow the actual signal waveform but only the envelope (provided the fre- quency is not too low) so distortion produced by the circuit is fairly small. The thermal inertia of the filament means, however, that the circuit cannot respond quickly to sudden increases in signal, so that associated circuitry may be overloaded whilst the lamp resistance is changing. Also the range of this type of compressor is limited. An alternative solution would seem to be the use of a voltage-dependent resistor (VDR) as in figure 5. This has a voltage versus current curve which is approximately the inverse of that of the lamp, so it is included in the lower limb of the attenuator. As the signal is increased the resistance of the VDR decreases so a smaller proportion of the signal appears across it. The response time of a VDR is quite fast so that it will follow sudden increases in signal amplitude, but unfortunately it can also follow the signal waveform so that instead of compressing the envelope amplitude whilst preserving the wave- shape it simply ‘rounds off the signal peaks thus introducing distortion. None- theless, in certain applications where distortion can be tolerated, such as amateur radio transmitters or intercoms, it does have its uses. It thus appears that the compressor designer is caught between two stools. A slow-acting device will cause little distortion on sustained large signals, but will not react sufficiently quickly to prevent momentary overloads of the equipment, whereas a fast-acting com- pressor will react in time to prevent overload, but will of itself introduce distortion. Here, however, an unusual aural phenomenon comes to the de- signer’s aid. The ear is incapable of detecting even large amounts of distor- tion in transients, so that if a fast- acting compressor is applied to a sudden increase in signal it will prevent gross overloading of the system whilst the distortion it introduces will be unno- ticed. Once the compressor has limited the signal, however, the ear can detect the distortion it introduces, so on sustained loud passages the slow re- sponse of the lamp-type compressor is required. In fact what is required is a compressor with a fast attack and slow decay characteristic. 442 — elektor april 1975 The characteristics of various types of directly by the signal on which they compressor are given in figure 6. The operate, but for a device with different triangular waveform was used to show attack and decay time constants it is how distortion is caused by a fast- necessary to turn to active circuits. In acting compressor. figure 7 the signal passes through the The discussion has so far been confined input stage and into a voltage-controlled to passive devices that are controlled attenuator. The output voltage is taken parts list: Cl2.Cl3 = 47#i. 10 V Rl,R4,RlO,Rl2 = 10 k Ti,T 3 to Tg = BC 109C Di = zener diode 2,7 V T 2 = BC 179C D 2 to D5 - germanium diode R 3 = 4k7 R 5 = 220 fi matched pairs A A 1 19 Dg to Dg - silicon diode 1N914 R6- R 17. R 20. R 26 = 22 k capacitors: R 7 = 1 k Ci = 100 n R 8- R 15- R 16 “ 330 k C2.Cn = 1 n. 10 V Rn = 270 k C 3 = 180 p R 13- R 14- R 25 = 3k3 C 4 - 100 H, 16 V for V b - 9 Volt: R 18 ,R 19 = 270 £2 R 24 “ 47 k C5,Cg,Cio “ 560 n and R 23 = 1k8 R 27 = 120k C 6 = 100p, 4 V for V b = 12 Volt: Rl8.Rl9 = 330£2 Pi = preset 22 k C 7 .C 8 = 2,2 H, 10 V and R 23 = 1k5 Figure 8. An LDR used in a voltage-controlled attenuator. This circuit suffers from slow response due to the inertia of the lamp and LDR. Figure 9. An r.f. carrier type of compressor. The filter eliminates harmonic distortion of the carrier caused by the attenuator and also eliminates control-voltage noise. Figure 10. Voltage-current curve of a diode and circuit of a simple diode attenuator. Figure 11. Balanced type of diode attenuator eliminates control-voltage noise which appears in common mode. Figure 12. The circuit of the final compressor design. Figure 13. The printed circuit board and component layout of the compressor. elektor april 1975 - 443 from the output of the attenuator and is also fed to a peak detector which rectifies the signal. The rectified voltage charges up the capacitor C via the potential divider consisting of Rj and R 2 . The time constant is The voltage on C increases the attenu- ation of the voltage-controlled attenu- ator as the signal increases. If Rj is small C charges up quickly but since the dis- charge path for C is via R 2 only, the decay time constant can be made as large as desired so that the voltage on C will not follow the signal waveform. as otherwise the variations in control voltage with varying signal levels will appear as spurious noise at the output. One way of achieving this would be by using a light-dependent resistor (LDR) as the lower limb of the attenuator, as in figure 8. This would be controlled by a lamp driven from the control voltage. Unfortunately problems arise due to the slow response of both the lamp and the LDR. Another rather elegant solution is to amplitude-modulate the signal onto a carrier and to vary the modulation depth by a voltage-controlled amplifier stage (figure 9). The compressed modu- lated signal is then filtered to remove control voltage noise and distortion (mainly second harmonic) and is then demodulated, resulting in a ‘clean’ compressed signal. Intermodulation distortion can still occur, but this can be minimised by proper circuit design. The design chosen for the final circuit to be described was a diode attenuator. In its simplest form (figure 10) it suffers from two disadvantages. 1. The signal voltage will itself vary the attenuation as with a VDR thus causing distortion. 2. The control voltage will appear at the output superimposed on the signal thus producing spurious noise. The first problem may be overcome by making the signal small compared with the control voltage so that it has little effect. The second may be prevented by using a balanced attenuator of four diodes as in figure 1 1 . The signal appears differentially at the input of the differ- ential amplifier and is therefore ampli- fied. The control voltage, however, appears in common mode and is there- fore rejected. The voltage-controlled attenuator Whilst the derivation of a control voltage from the signal is a relatively simple matter the design of a suitable voltage- controlled attenuator is another matter. Ideally the attenuator should be electri- cally isolated from the control voltage The Final Circuit Figure 12 shows the circuit of l 444 — elektor april 1975 disc preamp Compressor Specrficatio Input impedance 180 k Output impedance 25 k (do not load with less than 100 k) minimum Gain with Pi at 60 (max. input voltage = 1 V) maximum Maximum (compressed) 150 (max. input voltage = 30 mV) output voltage Maximum distortion (gain ■ 60) a. below compression 500 mV threshold b. at maximum (1 V) 0,4% Maximum control current through diode 5% bridge 350 pA Power consumption 10 mA at 9 V disc preamp A preamplifier-equaliser for magnetic pickup cartridges has to meet quite exacting requirements. Values for gain, noise level and maximum input voltage which will guarantee trouble-free operation under all conditions are not so easy to achieve. The well-known two-transistor configuration, operating from a 12 ... 18 V supply, invariably falls short on gain and overdrive-margin - unless it is designed for a low nominal output voltage (about 30 mV). An alternative approach is to make use of a good integrated amplifier. The design about to be described, which meets all the requirements, employs a SN 76131. An almost identical I.C. is the pA 739. compressor intended principally for speech applications. The circuit has an input stage with adjustable gain which is sensitive enough to be driven by a magnetic microphone. This is followed by a phase splitter which produces two antiphase signals to feed into the differential stage T4, T 5 . The com- pressed output is taken from the collector of T4 which should not be loaded with anything less than 1 00 k as this would upset the circuit oper- ation. A class B-type stage T7, Tg drives the peak detector Dg , Cu . The control voltage appearing on Cu is buffered by the emitter follower T« and is fed to the diode bridge D 2 ... D s . Di is a threshold control which determines the point at which compression starts. is simply a constant current source for the 'j differential pair. 1 1 The board and component layout for I the compressor are given in figure 13 j and the performance figures in the table. I I At first sight it may seem that the I distortion with the compressor oper- [I ating is rather high but compared with the distortion when an amplifier is over- I loaded it is minimal. Applications of the compressor J This compressor is sure to find a whole J host of applications. It can be used to 1 control the recording level in a tape L- recorder to prevent overloading of the j tape. It can be used in amateur radio I installations to achieve the largest || possible modulation without over- modulating so that maximum range can be achieved. It can be used in a car radio so that quiet passages may be heard above the engine noise without loud passages being unbearable. The I range of applications is limited only by ’ the ingenuity of the constructor - remember, a compressor rules the waves I I (somewhat straighter than they were f originally!). Bibliography: Electronic Engineering, January 1973. I Radio Elektronika, January 1959. To make optimum use of the possi- bilities for groove-modulation, gramo- phone records are cut with low audio frequencies attenuated and high audio frequencies boosted (with respect to 1 kHz). To simplify playback equalis- ation, a single weighting curve has been standardised throughout the world - the IEC disc-cutting character- istic. (This curve originated as the RIAA standard: Record Industry Association of America). The disc-cutting engineer arranges for a ‘0 dB standard (reference)level’ in the taped programme to produce a stylus tip-velocity about 14 dB below the ‘safe’ drive-level, to provide headroom for instantaneous signal peaks. 0 dB standard level (corresponding roughly to the average level in loud passages) is typi- cally 39 mm/sec tip peak velocity at 1 kHz. Standard level on carrier-channel discs (CD4 and UD4) is lower, about 22 mm/s. Experience indicates that wide-band cartridges suitable for carrier discs deliver 70 ... 140 pW for each mm/sec of tip velocity. The usual ‘hifi’ cartridges deliver about 6 dB more. (Note that sensitivity specifications are usually given in RMS millivolts per peak centi- metre per second). So the input to the preamplifier at standard level 1 KHz will be about 1 ... 10 mV peak. What are the consequences of all this for the preamplifier? Suppose it is the intention that the output voltage at standard level be about 100 mV RMS with the lowest-output cartridge. The closed-loop gain must therefore be 100 at 1 KHz. Now allow 20 dB of extra gain for IEC equalisation at the lowest frequencies, not including 20 dB of negative feedback (which should reasonably be maintained at the ‘low end’). This tots up to an open-loop gain of at least 80 dB! Ten thousand times. That seems to eliminate the two- transistor configuration. The SN 76131 integrated circuit, with the chosen lag compensation, has a typical open-loop response according to the upper dashed curve in figure 1 . The Figure 1. The desired closed-loop gain curve follows the IEC (RIAA) disc equalisation characteristic, with a mid-band gain (1 KHzl of 40 dB (heavy line). The open-loop gain must be at least 20 dB greater; the SN 76131. with the chosen lag compensation, provides this with a margin of about 10 dB (upper dashed curve). Figure 2. The heavy line is an estimated contour for the highest voltage delivered to the preamplifier by a high-output dynamic cartridge. The preamplifier cannot be over driven by the highest input voltage; the upper dashed line is the overdrive threshold for the disc-preamplifier with SN 76131. This cl- the maximum-input contour by approximat 10 dB. Figure 3. The maximum RMS output l~ produced by the preamp when used with high-output cartridge follows the thick ; tour. The dashed line indicates the maxim output capability. The safety margin is h once again about 10 dB. etektor april 1975 - 445 lower dashed curve indicates the mini- mum requirement (80 dB at the low end, reducing as the closed-loop gain - i.e. the bold line in figure 1 - falls according to the IEC curve). The con- clusion is that there is about 10 dB of open-loop gain to spare at all fre- quencies, which will accomodate 1C- tolerances etc. Overdriving the input To find the maximum input voltage which can occur, one must start with the highest-output cartridge. This will deliver, as shown earlier, about 5 . . . 10 mV peak at standard level. The maximum level encountered on the disc is nominally +14 dB relative to standard level. This indicates a nominal maximum input voltage of 25 . . . 50 mV. (At 1 KHz of course). It is clearly advisable to regard this figure, with due respect, as nominal. One might encoun- ter a cartridge with still higher output or some disc manufacturer may fully exploit tracing-compensation, to cut a clean signal at more than +14 dB . . . The absolute limit (set by ‘slope-over- load’ at the inner radius of LP discs) is presently about 350 mm/s (+18 dB) - but a 33 disc also has outer grooves and they can be cut at a level 6 dB higher. This means that in theory the maximum output level for the highest output cartridge is about 200 mV! With the circuit arrangement given, the SN 76131 will accept 80 mV at the input (thick dashed line in figure 2). The same figure can be used to estimate the effect of amplifier noise. The wide- band noise level, referred to the SN 76131 input, is 2 /iV (RMS). This is —68 dB in the figure (0 dB = 5 mV RMS). For the least sensitive cartridge, this noise level is -54 dB relative to standard level for CD-4 or UD-4 discs. Assuming maximum signal level to be +14 dB the overall S/N ratio is (for this worst case) 68 dB. Manufacturers esti- mate that the S/N ratio possible with a first-rate LP pressing is about 70 dB. Conclusion: pass. Figure 2 can be used once more to determine the hum-level requirements. The IEC bass-lift now aggravates matters: to achieve a hum level 60 dB below standard level, with a fairly high-output cartridge (5 mV RMS at 1 KHz), it becomes necessary to keep the hum voltage at the input below 1 pV! This can be achieved, in general, by providing good screening for the input circuit and for the preamplifier itself (signal-return inside the cable-screen, the latter bonded to signal-earth at the amplifier end only), and by properly smoothing (preferably regulating) the DC supply. The sensitivity of the SN 76131 to interference on the DC supply rail is quoted — under operating conditions rather different to the above - as 50 mV/V. (i.e. 50 /rV apparent input for each volt of supply disturbance). To achieve the 1 £iV hum level just men- tioned means keeping supply ripple below 20 mV. A simple active circuit will readily meet this requirement; 446 — elektor april 1975 disc preamp simple smoothing of a ‘raw’ DC supply would probably be inadequate or too expensive (or both!). Clipping at the output The requirement that the input circuit is not overdriven will not by itself quarantee that the amplifier as a whole operates within limits. The output circuit can still ‘run out of voltage or current swing. Taking the combination of a sensitive car- tridge and the maximum disc modulation likely to be encountered, one can esti- mate the highest level of output signal that the preamplifier will have to deliver. This can be done by combining the closed-loop gain characteristic (figure 1 , thick line) with the maximum cartridge output contour (thick line in figure 2). The result is shown in figure 3 (thick line). The conclusion is that the voltage swing at the output can be as high as 2.5 V RMS (7 V p-p). The clipping level for the SN 76131 depends on the supply voltage and on the load impedance. The case of V + = 30 and Rl = 10 K, where the 1C can deliver about 7 V RMS, is shown dashed in figure 3. This reserve should take care of all eventualities. If one considers a brink- of-disaster capability of 3 V RMS, then the combinations 18 V/5 K, 14 V/10 K and even V+ = 12 (at Rl = 50 K) are in order. Even under these conditions, current clipping due to the load of the feedback network on the output (at the highest audio frequencies) and slew-rate limiting (due to the early open-loop rolloff) are not expected to occur. Integrated circuit The circuit was designed around the specified SN 76131 by Texas Instru- Figure 4. The pinning of the IC's SN 76131, TBA 231, TCA 590C, [lA 739C and LM 1303 (figure 4a) it identical. The internal circuit diagram (figure 4b) however only applies to the SN 76131. Figure 5. The circuit diagram of the equali- preamplifier. An integrated voltage regulator, when required, can be connected between the points A and B (see text). Figure 6. PC board and component layo for the equaliser-preamplifier. All exte~ connections are made to one edge of the PC5 board, so that it can be used as plu* module in a complete control amplifier. Figure 7. Illustration of the preamplifier bo- as plug-in module. fcc preamp Table 1. The most important specifications o the SN 76131 and / JA 739C. II r™- 36 V ±5 V 500 mW 1 V nilt swinq 1 ... 26 V* Open loop gain typ 18000* Open loop gain min 6500* z in *VP- 150 KS2* Z in min. 37 Kf2* Z out (1 KHz) Crosstalk (10 KHz) -140 dB* 1 * These values apply for R|_ = 50Kft V+ = 30 V; r merits. According to the maker’s data [sheets, the Fairchild pA 739C and the TBA 23 1 are almost identical and perform well in the circuit. The IC’s are pin compatible (see 4a). Two other IC’s with the same are the Philips TCA 590 C and Table 2. Main specifications of the disc pre- amplifier described here. the LM 1303 by National Semiconduc- tors. This last device has lower specifi- cations for gain, noise and drive level - it will probably work acceptably in the preamplifier, but we have not checked this. The internal circuit of the SN 76131 (and the TBA 23 1 ) is given in figure 4b. Except for the output transistor, the M 739C is identical. Table 1 lists the most important characteristics of the device. The TCA 590C has an additional class B output stage, while the LM 1303 circuit dispenses with the stabilising diodes and with the current sinks for the second long-tail pairs. The external circuit Figure 5 gives the complete circuit dia- gram of the equaliser-preamplifier. The open-loop response is set up by C 4 /C s /R 3 and C11/C12/R11; it follows the appro- priate dashed curve in figure 1. The IEC correction networks are R1/R2/R4/C1/C3 and R12/R13/R14/C13/C15. R s and R] 0 take care of the DC biassing. With the values given, the correction obtained using 5% components is within 1 dB of the IEC (RIAA) standard. The input blocking capacitors C 7 and C9 should not be replaced by larger values or by electrolytics. This could lead to undesirable switch-on phenomena (‘plop’ or even momentary oscillation). The values given will not affect the bass response (which is 1 dB down at 20 Hz). It has already been pointed out that the supply ripple must be well filtered. A typical regulated supply will meet the requirements, but a ‘raw’ supply followed by resistor-electrolytic filter will usually cause too much hum. In this case one can use an IC voltage regulator which will deliver 24 ... 30 V at 15 mA (or more), e.g. the Fairchild jiA 78M24HC. The printed circuit board (figure 6) has a position for this regulator. If such a device is not to be used, the points A and B should be bridged. To simplify assembly, all external con- nections have been placed at one edge of the PC board, using standard grid- spacing. A control amplifier which will be published at a later date has a PC board designed to accomodate the disc preamplifier as a plug-in module (figure 7). Table 2, in conclusion, summarises the most important specifications of the equaliser-preamplifier for disc records. Lit.: Texas Instruments data sheets for SN 76131. elektor april 1975 a/d converter The convert a voltage C^IIIhkSI ■ B5S to a frequency such that the V frequency is accurately proportional to the voltage is one which arises in many different electronic systems. Some digital voltmeters use this principle. The voltage to be measured is converted to a proportionate frequency, which is then measured by a conventional counter circuit, and the result displayed digitally. In other cases, the requirement is to have a reading of a voltage existing some distance away. In this case the long cables, with their appreciable DC resistance, produce a voltage drop if any current is taken by the measuring instrument, and errors result. If however the information is carried over the cables as a frequency, although the amplitude may fall the frequency will not change. Increasing use of digital computors, digital logic IC's, digital displays, etc., produces many more applications. A previous design for a convertor circuit gave reasonable performance. However, further work produced several relatively minor changes which improved both linearity and temperature stability, resulting in the circuit described below. It is relatively difficult to convert voltage to frequency in a direct manner, if good linearity is to be maintained. However, the reverse operation, frequency to volt- age, is much easier. With this in mind, the method used here is firstly to convert voltage to frequency in a circuit which in isolation would not be very linear. The output frequency is however then con- verted back to voltage in another circuit (which this time is highly linear) and the output voltage used in a negative feed- back loop path so as to linearise the whole system. The overall linearity of the system will then approach that of the frequency-voltage convertor, pro- vided the feedback loop gain is high. Figure 1 shows the block diagram. The high gain differential comparator (A) accepts the input voltage and compares 1 frequency/ differential amplifier u \® voltage/ * rTn 1 -o li . / © | it with the feedback voltage. The voltage- frequency convertor (B), which can be relatively non-linear, is driven from (AX Its output provides the system output, and also drives the highly linear fre- quency-voltage feedback stage (C). j The Basic Circuit This circuit is shown in figure 2. A* IC type 741 is used in conventioni manner as the differential comparator The system input voltage is applied U the non-inverting input pin 3, and thf feedback voltage to pin 2. The outpu from pin 6 then regulates the frequenq of the next stage in such a way that th« two inputs remain almost identic^ Capacitor Ci provides AC negative feed back, to prevent appreciable AC sigr appearing at pin 6. The IC type 709, with the other cc ponents in the dotted line box (B) afl figure 2, together form a square wav oscillator. Consider first the case wheJ Q'” all ©_ a/d converter elektor april 1975 - 449 Figure 1. Block diagram of the complete Figure 2. The basic circuit. Figure 3. Showing extra transistor used to improve zero error. Figure 4. The final circuit of the convertor. The conversion factor is set by Rig. there is a negative voltage on C 3 . Of the two differential inputs of the 709 (which is a differential amplifier), pin 2 will be more negative than pin 3, due to current in R 6 from the negative rail. The output at pin 6 will therefore be hard positive, holding T 2 in saturation. Provided the 741 output is sufficiently positive, C 3 will charge up via R 3 until it raises pin 2 of the 709 to slightly above pin 3. As soon as this happens, the 709 output pin 6 rapidly goes negative, thus cutting off T 2 and allowing the system output to rise to a voltage determined at about 8.2 V by D 6 . Immediately, the voltage pin 2 is drawn even more positive by current through R 7 . Thus the action is regenerative. This condition now remains for a time 1 1 , which is determined by the values of R 7 , Rg, R 9 , Rio, and C 4 . (The value of ti is not affected by the voltage on C 3 , because as soon as 709 pin 6 goes negative, C 3 is driven rapidly negative via R 14 , D 4 , and Ri S .) C 4 is charged positively via R 9 and Ri 0 until pin 3 becomes more positive than pin 2, at which point the 709 output reverts, T 2 is once more turned on, and the cycle starts again with C 3 charging up. The result is a series of rectangular pulses at the output, whose width is ti and whose amplitude is constant (at the Zener voltage). Their PRF will however be determined by the time taken to charge C 3 , and hence by the 74 1 output voltage, so that overall the 741 input voltage con- trols frequency. D s is included to protect T 2 from excessive reverse voltage on its base. The circuit including Ti , R u , R 12 , Ri 3 , D 3 , and C s is put in to discharge C 4 to zero at the end of the period t, , and is driven by the positive-going step change from the 709 pin 6. The regen- erative action, via T 2 , is speeded up by capacitor C6 . The frequency-voltage convertor is sur- prisingly simple, comprising only R 4 and C 2 ! It merely smooths out the AC com- ponent of the rectangular wave, leaving on C 2 the DC component, whose value is exactly proportional to the PRF, or frequency. Improved Performance It is a defect of the above system that the saturation voltage of T 2 (i.e. its collector-emitter voltage when turned hard on) is not exactly zero, but can be something around 40 mV. Worse still, this value varies with temperature. This has the effect of producing a zero error con- sidered at the system input, so that with short circuit (i.e. zero) input the output frequency cannot always be set to zero by R 23 . This can be balanced out as shown in figure 3. An extra transistor T 3 is used, which is biassed by R 19 and R 20 so that it is permanently in saturation. The de- gree of saturation is governed by R 20 , and can be adjusted so that the saturation voltage equals that which occurs in T 2 . This voltage is applied to pin 2 of the 74 1 , via R 2) , and since the value of R 21 almost equals the value of ( R j + R 4 ), the voltage at pin 2 is exactly zero when T 2 is saturated. An extra spin-off from this arrangement is that temperature variations in the satu- ration voltages of T 2 and T 3 will approximately track each other, and so be balanced out. This modification is shown in the revised circuit of figure 4, together with several other changes as follows: (a) R 2 is reduced to 10 k. (b) C 2 is increased to lp in order to improve accuracy for low input voltage levels, where frequency is of course low, and a longer time constant is desirable. (c) C 4 is increased to 10 n, improving linearity at high input voltage levels. 450 — elektor april 1975 a/d converter S iiiiiSSsiiiliiiii '• i i (d) Trimmer R )0 is increased to 22 k, so that despite the increased C4 value, ratios of up to 10 KHz/V can still be set up. Setting-up procedure The sequence is as follows. The collectors ofT 2 and T 3 , and also the voltage input, are temporarily shorted to earth. The zero offset pot R 23 is then set to give zero volts at the 741 output. This adjust- ment is easier if a 100 k resistor is temporarily strapped across C, . The shorts across T 2 and T 3 , and the 100 k resistor can now be removed. The pot R 20 is set up for zero output frequency with zero input voltage. It should be remembered that since a negative frequency is meaningless (!) this setting should be approached by lowering the input voltage from positive towards zero, and observing the frequency to decrease and become zero simultaneous- ly with input voltage. In some cases it may be necessary to alter the value of R19, to compensate for unusual current gain values encountered in T 3 . In the same way, R 2 o can be increased to — say - 470 k. The next stage is to set the voltage- frequency conversion factor. The short Figure 5. Test results. Figure 6. PCB layout and component pos- Figure 7. Photograph of setting-up procedure. Here the DVM has accuracy better than 0.035%, and the frequency counter better than 0.01%. circuit on the voltage input (obviously!) should be removed, and a source of exactly 1 V connected. To achieve the best accuracy of which the circuit is capable, this value should be set up with a digital voltmeter, or other instrument, having better than 0.1% accuracy. The output frequency can then be monitored on a counter and set up using the Pot R| to the value desired. The design centre] value for this circuit is 1 0 KHz/V, but of] course other values can be set up required. Performance Details Figure 5 gives graphs of the circi performance after setting up as describ above. The error in volts, over the whole] input range 0-1 V is less than 1.5 mV.] Further, the relative error over the rai 7 mV to 2.5 V is less than 1% of readi The circuit was also tested without R ; connected. With R 20 it was set up wil short circuit input to zero Hz, and the with Rio to give 10 KHz for 1 V input| It is to be expected in this case that bol linearity and temperature stability woul be worse. Despite this, the accuracy « ± 1% over the range 0.1-1 V was mail tained. Best temperature stability will be ol tained by choosing C 2 , C 3 , and C4 car fully, poly-carbonate types being 11 commended, (e.g. Siemens MKM). Lii earity can be further improved by usii a faster OP-AMP in place of the 709, an by replacing R4 by a constant-currei source. However, these sophisticatiol are only worthwhile if really accural test gear is available for setting up. | Z y • fl ■ * 1 a inn led-displays elektor april 1975-451 led displays In response to numerous requests from readers we publish this LED-display chart to enable constructors to find their way through the jungle of seven- segment display data and to choose alternative displays to those specified in Elektor projects. There are literally dozens of different seven-segment LED-displays currently available and it would be prohibitively expensive to specify and test every suit- able alternative display for Elektor pro- jects. This guide is intended to enable the home constructor to choose such alterna- tives himself. This guide is confined to displays with dual-in-line pin configuration and common-anode connection as this is the most popular format and these displays can be driven by the common 7447 TTL decoder driver or interfaced with MOS devices by single-transistor buffer stages The guide is divided into three sections: 1 . The chart proper. 2. Condensed data on the devices giving pin connections and important performance parameters. 3. Hints on choosing devices and calcu- lating current limiting resistors etc. The Display chart This is used in a similar fashion to a mileage chart in a road atlas. The required device is first located in the diagonal list of type numbers (they run alphabetically by manufacturer, top left to bottom right). The proposed alterna- tive is similarly located and if the box where the corresponding row and column cross contains a circle than the devices are direct replacements for each other. If the box contains a P they are pin com- patible but the performance data should be checked to see if a substitution may be made. For instance, devices of differ- ent colours may be pin compatible. Figure 1. The LED-Display chart which may be used to find pin-compatible and direct replacement displays. led-displays elektor april 1975 — 453 The Data The data has been extracted from the relevant manufacturers’ data sheets and presented in tabular form. An expla- nation of the symbols used is also given. Where a particular box in the table con- tains a dash this means either that the parameter is not specified or that the units are not the same as the units used in the table. For example, some manu- facturers specify luminous intensity in millicandelas whereas others specify brightness in foot-lamberts. When using the data to choose alterna- tive devices it is important to check the physical dimensions of the device. Many devices have bodies that overhang the pins, so when using a p.c. board on which several devices are stacked close together check that there is sufficient space to accommodate the width of the device you propose to use. Some devices have bodies which are not symmetrical about the pins but are offset to one side and these have been excluded from this guide. Choosing Seven-Segment Displays When choosing alternative devices to those specified in a project, note the following points: 1. Devices that are not directly pin- compatible may often be used with slight modification. Many devices have similar connections for the segment cathodes and vary only in the connections to anode or decimal point. The most fre- quent connection for the common anode is pin 14. Some devices have additional anode connections, notably pins 3 and 9. In some cases the various anode pins are interconnected inside the package and are therefore redundant, in other cases the different anode pins are connected only to certain segments and must be connected externally. It is a simple matter to check which is the case. Con- nect one of the anode pins to a +5 V supply via a suitable limiting resistor and ground each of the segment pins and the decimal point pin in turn. If all the segments and the decimal point light then the anode pins are redundant and only one of them need be used. When using a device with redundant anodes in a circuit board with a single anode connection simply cut off the unused anode pins. When using a device with multiple non-redundant anode con- nections it is necessary to bend the extra pins inwards and link them to the pin used for the common anode connection to the circuit board. Devices with fewer anode connections than the device for which a board was originally intended present no problems, provided that the pins which will go into anode connection holes on the board are NC or may be cut off. 2. Some devices are available in versions with left- or right-hand decimal point and are identical but for the decimal point connection. In applications where the decimal point is not required (for example clocks) the pin may be cut off if necessary. 3. Having established that a device is, or Figure 2. Table of pin connections and performance data for LED-Displays. Figure 3. General appearance of a seven- segment LED-Dispiay showing letter desig- nation of segments and decimal points. Figure 4. Relationship between light wave- length and perceived colour. can be made, compatible with the board, the next thing to determine is whether the opto-electronic characteristics of the device are suitable. One of the most fre- quent of readers’ queries concerns the substitution of displays of different colours to the ones originally specified. Provided the electrical characteristics are suitable this is perfectly acceptable but it must be remembered that yellow and green devices are often less efficient than red ones. This is particularly true of older designs of device which are often available on the amateur market. Yellow devices are the least efficient of all. This effect is fortunately offset to some extent by the fact that the eye is more sensitive to yellow and green than it is to red light, so although yellow and green displays are often less bright than red displays operated at the same current the apparent brightness is not much less. Nevertheless the difference is often noticeable. 4. To ensure long device life it is advis- able to operate displays at or below the current specified in the If column. As this may involve recalculation of the cathode series resistors the formula is given below _ Vb ~ Vfe If Rk = - (Vb is the supply voltage; Vf s and If can be found from the table). A similar calculation may be performed for the decimal point cathode resistor by substituting Vfa for Vf s . When displays are used in a multiplexed (strobed) mode then they are only on for one nth of the time, where n is the num- ber of displays being multiplexed. Conse- quently, to maintain the same brightness as if they were being driven continuously they will need to be supplied with n times the current for the time they are on. Most displays can be strobed at sev- eral times the continuous forward cur- rent I av . The formula for calculating the segment cathode resistors then becomes: - Vb~ Vfs nl f When operating with low supply voltages (e.g. TTL 5 V) it is advisable to subtract the saturation voltages of any transistors used in the multiplex drive circuitry from the supply voltage when calculating Rk- Of course, where it is desired to increase the current when using an alternative display it is necessary to ensure that the circuit can supply the extra current. 454 - elektor april 1975 modulation systems part 2 modulation systems As already announced in Part I, this instalment deals with Carrier Position Modulation (CPM), as well as with frequency and phase modulation. Frequency modulation has proved to be the best system for VHF broadcasting, while CPM is most suitable for speech transmission. Carrier Position Modulation (CPM) When a speech clipper is used, two questions arise: 1. What can be gained by using this system? 2. To what extent is intelligibility affected? Experiments with HF clipping on SSB signals have demonstrated that intelli- gibility remains good even with infinite limiting, while the average power is increased by about 10 dB. If pre- emphasis is provided ahead of the LF chain, a further improvement in intelli- gibility results. Even with infinite limiting of an SSB signal there are still some variations in its amplitude, since the rapid phase jumps of SSB signals give rise to fre- quency components outside the trans- mitted frequency band. Since these components are filtered out of the (constant) RF signal, the resultant trans- mitted signal must contain amplitude modulation. If an SSB signal is to be purged of all amplitude variation, further signal pro- cessing is needed, and a PLL circuit happens to be suitable for this. Figure 16 shows the block diagram of an arrange- ment for producing CPM signals. An SSB signal is produced from a pre- emphasised LF input, and after this signal has been limited it is fed to a PLL circuit. The VCO in this circuit will oscillate at the same frequency as the SSB carrier, but without any amplitude variations. Component values in the PLL are chosen to make it unable to follow rapid phase jumps in the SSB signal, so that the bandwidth of the CPM signal is not much greater than that of the original SSB-signal. Always provided that care is taken to maintain intelligibility, remarkably high efficiency can be achieved with CPM. Figure 17 shows the relationship be- tween intelligibility and receiver input voltage for different modulation sys- tems. These are based on tests with sequences of unrelated words, and on the use of an IF section of the most suitable form for each system. For the same degree of intelligibility, the necessary input voltage with CPM is less than a third of what is needed with AM. This means that a CPM signal needs only one-tenth of the power needed for a 1 00%-modulated AM-signal, to cover the same distance. CPM thus offers higher efficiency for a given transmitter power. CPM has only been known for a com- paratively short time, and amateurs have experimented very successfully in this field. Arrangements based on the prin- ciple outlined in figure 16 are generally used, but this unfortunately has some disadvantages. The input signal to the balanced modulator exhibits amplitude variations depending, among other things, on the speaker and the speaker’s distance from the microphone. The SSB signal must have greater amplitude than the carrier injected through Pi, the purpose of which is to suppress noise originating from the limiter when there is no modulation. In producing CPM, the LF signal must be suitably processed to avoid these subjective effects, and as already indicated, this cannot be done with clipping alone. The results obtainable with rapid com- pression are almost the same as those which can be achieved with logarithmic amplification, so the latter method is preferable because of its simplicity. The block diagram of a CPM transmitter with LF-signal processing is given in figure 1 8. A frequency band of 400 Hz to 3400 Hz from the microphone is amplified logarithmically and fed to the balanced modulator. The LF signal now has only a small degree of amplitude variation, and it is therefore possible to inject a higher level of carrier than would be acceptable without logarithmic amplification. This results in a better signal-to-noise ratio for the transmitted signal. There is in addition another advantage to be gained from this configuration, namely that it can be used for phase modulation provided the level of carrier injection is sufficiently high. It can be shown that the PLL produces a phase- modulated signal if the balanced-modu- lator signal emerging from the filter is smaller than the injected carrier. The modulation index is a function of the quotient of these two voltages. After the VCO signal has been directly transposed to the desired transmission frequency, it can be brought up to the power required by a Class-C amplifier. CPM can be received in the same way as SSB, but as an unmodulated carrier component is available for part of the time, a PLL can be used. Since a CPM signal contains no amplitude infor- mation, amplitude limitation in the receiver raises no problems, and this offers a simple means of combating AM interference in mobile applications. By way of verification of the advantages of CPM, experiments were carried out with an Ultra Low Power transmitter using the principles of figure 18. A fre- quency of 27 MHz was used, and the Table 1. System Present application Efficiency Future applications AM Long-, medium- and shortwave broadcasting very low none DSSC SSB communications networks high very high present applications CPM exceptionally high communications networks and citizens' radio FM NB-FM PM high-quality broadcasting communications networks usually changed to FM with integrating networks high high moderate present applications continuing Long-, medium- and short- wave broadcasting current applications continuing modulation systems Figure 16. CPM (Carrier Position Modulation) is effected by processing an infinitely-dipped SSB signal using a PLL. Figure 17. Relationship between intelligibility and receiver input-signal voltage (50% intelli- gibility is considered inadequate). Note that the order of merit gives first place to narrow- band FM (NB - FM), and not SSB. Figure 18. Intelligibility can be considerably improved and lining-up simplified by the use of low-frequency signal processing. Figure 19. The spectral distribution of an FM signal can be expressed with Bessel functions. transmitter power was approx. 20 mW. In order to compensate, to some extent, for the unfavourable topographical con- ditions for VHF propagation - hilly country - the transmitter and its aerial were located on a floor of a block of flats 50 metres up. A loaded aerial rod was used, and the calculated efficiency of this combination was 30%, so that the ERP barely amounted to 6 mW. The receiver used for this experiment has a sensitivity of 0.1 /iV with a bandwidth of 3 kHz and was equipped with a PLL of the type shown in figure 10 (see Elektor no. 2). In spite of the transmitting aerial height, the optical horizon radius was a bare 7 km. Although reception within this area was subject to wide fluctuation, it was observed that the received signal did not drop below 0.2 #iV. The limit of receiver sensitivity was reached at a range of 10km, that is 3 km beyond the optical horizon. When the transmitter was switched over to phase modulation, reception at the optical horizon was observed to have already become insufficient for reliable communication. Frequency Modulation and Phase Modulation (FM and PM) When the frequency or the phase of a carrier is made to vary in accordance with information, this is known as fre- 456 — elektor april 1975 modulation systems quency modulation or phase modulation respectively. Both satisfy the relationship: V = V 0 sin(a)hf + m sinco[f t) (4) in the case of sinusoidal modulation. The difference between FM and PM lies in the modulation index m, which is defined for FM as: frequency deviation of the RF carrier with respect to the centre m _ frequency modulation frequency v and with phase modulation m is con- stant. The expression in (4) can be expanded to: V = V 0 J 0 m sin Whf t + Jim[sin(a>hf+wif)t + sinfcohf - wjf)t] + J2m[sin(o;hf + 2coif)t + sin(cohf - 2 o>if)t] + J3m[sin(cohf + 3wif)t - sin(cohf - 3coif)t] + . . . It can be seen from this that FM and PM generate a spectrum with infinite bandwidth. The term J n indicates a Bessel function of the nth order, whose magnitude decreases substantially as n increases. In practice both FM and PM can therefore be considered to have a finite bandwidth. Figure 19 shows the amplitudes of sideband components as a function of m. For FM broadcasting, a maximum fre- 21 input signal, RMS (pV) — quency deviation of 75 kHz and a maximum modulation frequency of 15 kHz were standardised at the outset. It follows from (5) that m = 5, and it can be read off from figure 19 that, with this modulation index, the relative amplitude of the seventh-order sideband is only 0.05. This can be neglected in most cases because maximum modu- lation does not occur at 15 kHz in practice. A rule-of-thumb formula, valid when m is unity or greater, is: B w = 2(m + l)fjf max, in which B w represents the —3 dB bandwidth, m is the modulation index at the maximum modulation frequency and fjf max is the maximum modulation frequency. The minimum bandwidth needed for mono FM then works out as: B^, = 180 kHz. For stereo FM the modu- lation index has been chosen, on com- patibility grounds, to enable the mono bandwidth to be used at the highest modulation frequency (53 kHz). The modulation index for the sub-canier signal conveying the stereo information can be shown to be 0.6 (as this is less than unity, the rule-of-thumb formula does not apply) which results in a 20 dB deterioration in signal-to-noise ratio. It can be derived from figure 19 that, with this low modulation index, the second-order sideband can be neglected in practice, as its relative amplitude is less than 0.05. The bandwidth required is then no greater than is needed for an AM system (2.fjf max). 24 Mod. Index =4 j | j jL Mod. Index ■ 12 jjyyj Mi Mod. Index -24 i i i i b 4» H frequency deviation Figure 20. Relationship between distance covered and signal-to-noise ratio for AM (1), FM with 20 kHz deviation (2) and FM w 75 kHz deviation (3). Figure 22. With low deviations, there is coi siderable improvement in intelligibility i low input-signal levels. Figure 23. These curves clearly show the superiority of narrow-band FM over AM of the same bandwidth. Figure 24. Spectra showing that bandwidth can be fully utilised by increasing the modu- lation index for lower frequencies. This is the case for FM, since the deviation is d< mined by the amplitude of the modulating Associated with this lower value of m is a drop in the maximum signal-to-noise ratio obtainable, and therefore in the suppression of both impulsive and adjacent-channel interference. One characteristic feature of FM is the threshold response: this means that the FM input signal strength must be above a certain value if it is to be usable. The threshold value goes down when a lower modulation index is used. This knowl- edge is based on research first carried out in the U.S.A. in the ‘thirties to determine whether AM of FM would be best for a reliable police radio network. Some of the results of these very extensive researches are reproduced in figures 20, 21, 22 and 23. The result of a terrain test is shown graphically in figure 20. In this case the transmitter location was fixed, while the receiver was mobile. Curve 1 is for AM, Curve 2 for FM with 20 kHz devi- ation and Curve 3 for FM with 75 kHz deviation. These curves show that a higher deviation is needed to give the high signal-to-noise ratio which hi-fi demands, but that a price has to be paid for this in terms of the maximum work- able range. Bearing in mind the 20-dB deterioration in signal-to-noise ratio with the present stereo system, it is of interest that stereo broadcasting using two separate trans- mission links, each with a deviation of only 20 kHz, would not only give a better signal-to-noise ratio, but would also offer a saving in overall bandwidth. For communication systems, used ex- clusively for speech transmission, a maximum modulation frequency of 3 kHz is adequate. Higher values, up to 5 kHz or 6 kHz, are used only when it is essential to transmit speech of very high intelligibility. Figure 2 1 shows a comparison between a system with 20 kHz deviation and one with 6 kHz deviation. It will be seen that the ultimate sensitivity of the narrow-band system is better by a factor of 2. In communications networks, a signal-to-noise ratio of approximately 12 dB is regarded as just usable. The amateur transmitters’ readability gradings are also often quoted in intelli- gibility tests. In figure 22 the amateurs’ intelligibility gradings are plotted against input signal for different values of deviation. The narrow-band system is quite usable with an input of 2 /iV, while the system with a 20 kHz deviation is un- readable with this input. Although limiting sensitivity is consider- ably better nowadays because of im- proved reception techniques, this has no effect on the relationships between limiting sensitivities with the various systems. A comparison between a narrow-band FM-system and an AM system with the same bandwidth is given in figure 23. The curves show the marked superior- ity of the FM system; this applies not only to intelligibility, but also to inter- ference suppression. This points to a possible alternative for medium waves which would at least reduce the chaos prevailing in this band. By re-engineering the present AM channels for narrow- band FM with a deviation of 4.5 kHz, a substantial improvement would be achieved. Anyone possessing a short-wave com- munications receiver equipped for narrow-band FM reception will find that, among others, a number of East European countries are radiating exper- imental narrow-band FM transmissions, particularly in the 25-m and 41-m broadcasting bands. These transmissions should be of particular service in shed- ding light on the effect of distortion caused by selective fading. This distor- tion seems to be considerably less with narrow-band FM than with AM and an envelope detector. As narrow-band FM is more compatible with AM than is SSB with a carrier, the change-over to narrow-band FM could take place gradually. This move is also advocated by the fact that, for the same signal-to- noise ratio, narrow-band FM would give a 70% saving in transmitter power. Particularly for narrow-band systems, I there is a great difference between I 26 " |— iDi-j r ' (FM + PM) -cm- -1 1- L-h-J 1 ) £ T LF phase- and frequency modulation. Sup- pose, for example, that the deviation with FM is 4.5 kHz and the modulation frequency is 450 Hz, giving a value of 10 for m. This will result in a large number of sidebands whose main energy content (J 8 in figure 19) is in the region of 9 kHz. With phase modulation m is constant so that, in the case of narrow-band PM, as a first approximation two sidebands will be produced, depending on the modulation frequency. The sidebands of AM and narrow-band PM are thus identical, and this makes it possible to receive PM with an SSB receiver. The efficiency of a communication system is highest when the available bandwidth is completely filled with information, and for this reason PM has a poorer signal-to-noise ratio than FM. This is illustrated in figure 24, where a higher modulation index is associated with lower modulation frequencies. Figure 25 shows a simple arrangement for producing a frequency-modulated oscillation by introducing a varicap into the LC circuit of a stable oscillator, so that the oscillator frequency will vary with modulation. This circuit can be changed over from FM to PM simply by feeding the modulation through an RC section whose cut off frequency is equal to the highest modulation fre- quency. As the building of an oscillator which satisfies Post Office stability requirements is not exactly a simple business, crystal-controlled oscillators are preferable. With these, however, direct modulation of the oscillator fre- quency is not possible, as the maximum deviation cannot be more than 200 parts per million. However a crystal oscillator giving phase and frequency modulation simultaneously may be used, and this can have the same overall effect for communications purposes. An example of such a circuit is given in figure 26. In many instances, however, the devi- ation obtained with this circuit will be too small, and the required deviation can then be obtained with a frequency- multiplier circuit. One of many possible variants of this circuit is shown in figure 27, and calls for a minimum number of stages. Crystal-controlled oscillators XTOi and XT0 2 oscillate at frequencies f] and f 2 . These frequencies are multiplied by n and (n + 1 ) respect- ively and then fed to a mixer stage, the output of which is tuned to: fout = ( n+ 1 ) f 2 -nfi. The two oscillators are modulated with opposite polarities, via a phase splitter, giving deviations of Afj and Af 2 respect- ively, so that the mixer output becomes: fout + ^h = (n+ l).(f 2 +Af 2 )-n(f, -Af,). This can be rearranged to give the value of the deviation fjj, i.e. : fh = (n+ 1)-Af 2 +nAfj. In the case where f j = f 2 and Afi = Af 2 , this gives: f h = (2n+ l)Afi, with a centre frequency: fout = • A practical value for n is three, as this can be effected with one stage of multiplication. This gives a seven-fold multiplication of the deviation, with an output frequency equal to that of the crystal oscillators. When FM signals are detected, imperfect demodulation causes divergences from theoretical values (e.g. for interference suppression), and these divergences increase as the bandwidth of the system is reduced. For this reason, special care should be devoted to the instrumen- tation of narrow-band FM systems, but unfortunately the opposite has been true in the past. Conclusion It has been shown that there are only two modulation systems offering high efficiency, namely FM and CPM. How- ever, since CPM conveys no information on amplitude, this system is only suit- able for speech transmission. AM is in every respect the worst system. Although it appears at first sight to offer economic advantages, closer study shows up the disadvantages of AM such as energy wastage, wavelength clutter and its contribution to the warming up of the ionosphere. FM has rightly been chosen for high- quality broadcasting, but even FM is marred by distortion and noise when new systems, which reduce the modu- lation index severely, are introduced. Stereo broadcasting with its information bandwidth of 53 kHz is a striking example of this, but it would seem that yet another step in the wrong direction is about to be taken with the introduc- tion of quadraphonic broadcasting. For a number of quadra systems now being discussed, a bandwidth of ‘only’ 76 kHz is needed. In view of the widespread operation of FM transmitter networks with a channel spacing of 100 kHz, it would be preferable to look for tech- niques which do not call for any increase in the present bandwidth. M Figure 27. With this arrangement, deviation can be increased without the increase in out- put frequency which occurs with direct multiplication. Modifications to Additions to I mprovements on Corrections in Circuits published Elektor Steam whistle In the p.c.b. layout for the steam whistle (Elektor 1, p. 58), the electrolytic ca- pacitors C 4 , C 7 and C 9 are shown with the wrong polarity. The negative con- nections of C 4 and C 7 should be con- nected to the negative supply line near the emitter of T 2 ; the positive con- nection of C 9 should be connected to the cathode of Di . The circuit diagram (figure 2) is correct. feedback PLL receiver TUP/TUN tester quadro in practice H/L logic probe jp-tun-dug-dus elektor april 1975 - 459 TUP TUP Tun Tun • UUE UUE UUE 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 above. For further information, see the article 'TUP- TUN-DUG-DUS' in Elektor 1, p. 9. Table 2. Various transistor types that meet the Table 4. Various diodes that meet the DUS or TUN specifications. DUG specifications. TUN BC 107 BC 108 BC 109 BC 147 BC 148 BC 149 BC 171 BC 172 BC 173 BC 182 BC 183 BC 184 BC 207 BC 208 BC 209 BC 237 BC 238 BC 239 BC 317 BC 318 BC 319 BC 347 BC 348 BC 349 BC 382 BC 383 BC 384 BC 407 BC 408 BC 409 BC 413 BC 414 BC 547 BC 548 BC 549 BC 582 BC 583 BC 584 Table 3. Various transistor types that meet the TUP specifications. TUP run DUE DUS DUS DUG BA 127 BA 217 BA 218 BA 221 BA 222 BA 317 BA 318 BAX 13 BAY61 1N914 1N4148 OA 85 OA 91 OA 95 AA 1161 Table 5. Minimum specifications for the BC107, -108, -109 and BC177, -178. -179 families (according to the Pro-Electron standard). Note that the BC179 does not necessarily meet the TUP specification Oc.max = 50 mA). NPN PNP BC 107 BC 177 BC 108 BC 178 BC 109 BC 179 v ce 0 45 V 45 V 20 V 25 V 20 V 20 V V eb 0 6 V 5 V 5 V 5 V 5 V 5 V •c 100 m A 100 mA 100 mA 100 mA 100 mA 50 mA p tot. 300 mW 300 mW 300 mW 300 mW 300 mW 300 mW f T 150 MHz 130 MHz mi n 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: a' (P. h fe ) = 125-260 B: a' = 240-500 C: 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. 460 - elektor april 1975 led-level I ; led-leoel In general, analogue pointer in- struments are used for level indicators. Another method of indicating amplitudes and power is to use LED's. The advantages of this system include higher resistance to shock, better legibility from greater distances and the fact that the response time is unaffected by the mech- anical time-constant of a conventional meter. Apart from a practical level meter additional circuits are discussed. The most important of these is a simple overload indicator. Figure 1 gives a simple circuit with which the voltage amplitude on the loud- speaker output of an amplifier can be converted into light intensity of lamp L, . The limiting resistor R, is necessary only if the lamp can be overdriven by the amplifier. Of course with a single supply rail amplifier the circuit of figure 1 must be connected after the loudspeaker out- put capacitor. Otherwise the lamp would be constantly fed from the d.c. mid-point voltage of the amplifier output stage. Lamp Lj must bum brightest at maxi- mum output power. This power is nor- mally limited by the supply voltage of the output amplifier. In most cases it can be said that the maximum output is obtained if the amplitude of the output voltage is about 2 volts less than the supply voltage (also in connection with increasing distortion). If, for example, the supply voltage of the amplifier is 24 volts, the maximum swing of the output voltage will then be about 22 volts peak-to-peak. The maximum RMS output voltage of the output stage (from the example) is half the peak-to-peak voltage divided by \/2. This is about 7.8 volts. The maxi- mum voltage of the lamp is 6 volts, so the surplus of 1.8 volts must drop across Rj. The resistance value of Ri can now be calculated by dividing the residual voltage (1.8 volts) by the 50mA which is the maximum current for the lamp. The level indicator Such a simple system can, at best, give only an approximate indication of out- put and its effectiveness depends on many factors such as ambient lighting and the eyesight of the individual user. A much better arrangement is to have a number of lamps or LEDs which light in sequence as the voltage is increased. This is the system used in the LED level indicator. The circuit is shown in figure 2. The input of the circuit is formed by potentio- meter P, with which the sensitivity is adjusted. The potentiometer is connected to the loudspeaker output of the ampli- fier. If the amplifier is fed asymmetrically (one supply voltage), potentiometer P! must be connected after the loudspeakei output capacitor. The circuit operates as follows: I- «imim mu mu mimcii; uuu r led-level Part* list with figure 4 Resistors: Rl = 1 M R 2 = 330k R3.R4= ’Ok R 5 - 270ft R 6 = Ik R7.R8 = 47fi Pi ■= 1 M, preset potentiometer Capacitors: Ci - 0.47/i C 2 = IOOjU/10 V (see text) C 3 - 1 00fj/35 V Semiconductors: t 1- t 2« t 3 " TUN (above Ufc - 20 V: BC107a) T 4 = 2N1613 T 5 = 2N2905 D 1 ,D 2 - DUS Figure 1. The simplest form of level indicator can be made with a lamp and a resistor. As the output voltage increases, the lamp will produce more light. The indication of such a system is not accurate, and for small voltages the lamp does not light. Figure 2. The LED level indicator fitted with ten LED's. Each time the output voltage in- creases by about 0.7 V an additional LED will light up. If the LED's are mounted in line horizontally or vertically) the result is a ''ther- mometer" type indication. The length of the track is an indication of the amplitude of the output. It is possible to use lamps instead of LED's. Depending on the type of lamp used, the load resistors Ri i up to and including R20 may be omitted. Figure 3. To obtain an indication at low out- put voltages the anode of diode Di of figure 2 must receive a bias voltage. This is done by means of an additional adjustment potentio- meter (P v ), resistor (R v ) and diode (D v ). Figure 4. If the level indicator must be driven from a high-output-impedance or low-voltage source a preamplifier circuit can be used. Its voltage amplification is 100 or more, depend- ing on the gain of T3. Figure 4a. This voltage doubler can replace diode Di (figure 2) if the indicator fails to give full deflection. The voltage doubler con- sists of two diodes (D, and D r ) and two capacitors (C, and C r ). The doubler can only be used if the meter has an independent supply. As appears from the diagram, the loudspeaker zero and level meter zero (minus terminal of Ci) are not D.C. connected. The output voltage of the amplifier arrives on diode Di via potentiometer Pj . This diode rectifies the signal positively. Via D, capacitor C, is charged. If the voltage across C, increases, there will come a point where T, conducts. If the voltage on Ci rises further, transistor T 2 will be driven into conduction via re- sistor R, . A resistor and LED are included in the collector of T 2 . When T 2 conducts, the LED lights. If the voltage on C, rises still further, transistor T 3 conducts be- cause its base is driven via diode D 2 and resistor R 2 . Now LED D, 2 will also light. As long as the voltage on capacitor C ! keeps rising, another diode in the chain D 2 ... D 2 i conducts. Each of the corre- sponding transistors (T 2 ... T„) and LEDs (Du ••• D 20 ) also conducts. When the emitter potential of T, is about 7 volts, all ten LED’s will be lit. elektor april 1975 — 46 1 If the LED’s are placed in line horizontal- ly or vertically the result is a light track whose length is proportional to the out- put amplitude of the amplifier. Potentio- meter P 2 in the emitter circuit of Ti serves to adjust and limit the current. The indicator responds rapidly to an increase of the output voltage of the amplifier. The decay time of the meter (light track) depends on the value of capacitor Cj. At a greater capacitor value the decay time becomes longer. At the indicated value for Ci the decay time is about 0.3 seconds. The circuit may also be fed from higher voltages. But then the values of R,, up to and including R 20 must be adapted. The proper values can be calculated if we assume that the supply voltage drops at least 1 .5 volts across a LED and that the current through the resistors is about elektor april 1975 40mA. (Ensure that the LED’s used will stand this current). If the supply voltage is more than 20 volts it is not possible to use a TUN. Up to a supply voltage of 40 volts the TUN’s can be replaced by BC107a or BC107b. Instead of LED’s ordinary incandescent lamps can be used. Their operating Figure 5. This overload indicator can be used universally. The input must be connected to the output of the amplifier before the output capacitor. Figure 6. If the level indicator must give an audiophysiologically corrected indication, this network may be connected between the loud- speaker output and the input of the meter. voltage can best be chosen to equal the supply voltage. In that case a load resistor (R n up to and including R 20 ) is not needed. A drawback of the circuit of figure 2 is that the first LED begins to conduct only after a bias has been built up. If this is unacceptable, the circuit can be pre- biased with a resistor, potentiometer, and diode. Figure 3 gives a detailed drawing of the input circuit of figure 2 with the additional components. The bias is ad- justed with potentiometer P v . Diode D v serves only to avoid extra loading of the positive-going loudspeaker signal. Level preamplifier If the level indicator must be connected to a point in the amplifier where there is not sufficient voltage (and power) to drive it, the circuit of figure 4 may be used. This circuit is inserted between the connecting point in the amplifier and I potentiometer Pj of figure 2. The input impedance of the circuit of I figure 4 is about 270 k. The voltage I amplification with P] at maximum is I 1 00 X or more. This depends on the gain j of transistor T 3 . The circuit of figure 4 I may be connected to supply voltages I between 1 2 volts and 40 volts. For sup- I plies higher than 20 volts the TUN’s I must be replaced by transistors which I can withstand this voltage (for example BC107). Furthermore, the operating I voltage of capacitor C 3 should be at least J equal to the supply voltage. If the supply voltage for the circuit of I figure 4 is less than 20 volts, the level I meter cannot be fully driven under nor- I mal conditions. To achieve this, diode D! | (from figure 2) must be replaced by a I voltage doubler, so that capacitor C] (of I figure 2) receives about twice the voltage I (see figure 4a). Overload indicator It can be quite handy if a power amplifier I is provided with a device that indicates I when the amplifier is overdriven: an I overload indicator. Figure 5 gives a prac- | tical example. The input is connected to I the output of the amplifier. Since we are now concerned with overdrive, the input I must be connected before the loud- I speaker elco. The threshold level of the overload I indicator may be adjusted by potentio- I meter P! . This adjustment must be such I that if a certain level is exceeded, the I fiA 741 switches, and produces a positive I voltage. This voltage drives transistor T i . The emitter circuit of Ti includes an I incandescent lamp or LED which then I 1 lights. I * The overload indicator of figure 5 can I also be used for higher voltages (up to I 37 volts). The value of resistor R 3 must I be increased in proportion with the 1 higher supply voltage. To ensure the I survival of the IC, the input voltage I should not be more than the supply I voltage. For this reason an extra resistor I (Rx) of 10 ... 22k in the input lead may I be needed. Physiological correction If the level indicator must give an audio- I physiologically corrected indication, the I network of figure 6 can be connected I oetween the meter and the loudspeaker I output. This network gives an attenua- I tion of about 4 X. If the input voltage is I then insufficient to drive the meter to I maximum indication, there are two pos- I sible solutions. The voltage doubler of I figure 4a can be used, or alternatively I _ the circuit of figure 4 can be connected I s between the correction network output ■ Li and the input of the level indicator. In I* that case potentiometer P! and the I ■ capacitors Ci and C 2 can be omitted- “ from the circuit of figure 4. With the audio-physiologically corrected level meter it is necessary to use an over- load indicator, because it is impossible to see when the amplifier is giving its peak power. M imiifMS'imPisfi; elektor april 1975 - 463 mHRKBT Quadrophonic cartridge from Elac A new range of pickup cartridges manufactured by Electro-acoustic GmbH of West Germany is now available in Great Britain. Illustrated is the ELAC STS 655- D4 cartridge, which is designed for playing quadrophonic carrier discs. It is fitted with a parabolically ground Shibata diamond stylus and will track at up to 50 kHz. The cartridge may also be used with normal stereo or matrix quadro- phonic discs. Elac cartridges range in price from £10- £49. Camouflaged Speakers For those who wish their Hi-Fi to be unobtrusive, the ECHONICA speakers from Japan may be the answer. Having the appearance of a picture in a frame only 1 Vi inches deep, these speakers are designed for wall mounting. The 'canvas’ of the picture is the loudspeaker diaphragm and a range of 60 pictures is available. The price is £47.00 a pair plus V.A.T. ‘I Low-cost 50-ohm Sweep/Function Generator A new sweep/function generator is available from Dana Electronics Ltd. The model 196 A offers sine, triangle, square, pulse, ramp and sweep waveforms over the range 0.1 Hz-1 MHz, in seven ranges. The generator will provide 10 V open-circuit or 20 V into a 50 ohm load. Attenuation up to 70 dB is provided in two 20 dB fixed steps and a 30 dB variable. An internal sweep generator will sweep the output frequency over up to three decades, with sweep rates from 1 mS to 10 S. A separate TTL compatible square- wave output is provided. Size is 187 x73 x 216 mm (7.5 x 2.9 x 8.6 inches) and it weighs less than 1 kg. Price is £195. Logic Probe A new TTL/DTL logic probe is available from Intercontinental Components Ltd. Readout is by four LED’s. H and L to indicate high or low logic states at the input and Q and Q, to indicate the state of a storage latch, which toggles on a positive transition at the input The probe derives its 45 mA supply current from the circuit under test and is reverse polarity protected. Probe input current is 2.4 mA max. and response time is 50 ns. The one-off price is £1 1.50. New family of low-power TTL devices National Semiconductor have announced the start of volume production of a new range of low-power TTL devices known as 54 LS/74 LS Low-Power Schottky or LPS. The first nine types, 74 LS00, 01, 03, 04, 10, 12, 20, 22, and 30 are now available in quantity. Suggested resale unit prices in lots of 100’s are £0.20 for all gates, except the 74LS04 which is priced at £0.22. It is anticipated that, by mid-1976, all of the popular circuits that are now in the standard 54/74 family will have been duplicated. The 54LS/74LS devices are claimed to have the best speed-to- power ratio of any high-speed logic family on the market Compared with standard TTL devices, low-power Schottky logic dissipates only one-fifth the amount of power (2 mW per gate) while making no sacrifice in operating speed. Low-power Schottky will replace most high-speed TTL logic, and can be used in some Schottky TTL applications, as well as standard TTL, since the LS series has dynamic characteristics that closely approximate those of the standard 54/74 TTL It is possible to remove a 7400 device and insert a 74 LS 00 device in its place and obtain the same speed with lower power consumption. Compact digital multimeter The ‘Danameter’ is an almost pocket-sized digital multimeter from Dana Electronics Ltd. The instrument is powered by a single 9 V transistor radio battery, which should last for up to a year of normal use, and has a 3!4 digit liquid crystal display that adjusts itself to ambient light levels. Sixteen ranges are selectable by means of a single, 18-position switch, with two positions for ‘off’ and battery test. The case is moulded in high-impact a-b.s. plastic and the manufacturers claim that the meter will survive bench-high drops and drastic electrical overloads such as 250 V on the ohms ranges. Ranges are 2, 20, 200 and 1000 D.C. and A.C. 20 p A, 2 mA, 200 mA and 2A D.C. 200 ohms, 20 k and 2 M. Dimension are 102 x 184 x 57 mm. (4 x 7.25 x 2.25 inches) and the weight is 0.45 kg (1 lb.). Price of the basic Danameter is £99.50. Versatile Multimeter A new multimeter is available from Metrawatt U.K. Ltd. The Unigor A42 Multimeter has a total of 30 measuring ranges for A.C./D.G current and voltage, and resistance. Ranges are: D.C volts, 60 mV- 12 kV, accuracy ± 1% F.S.D. A.C volts, 6 V- 12 kV, accuracy ± 1% F.S.D. /' N D.C amps, 60 flA- 30 A, accuracy ± 1% F.S.D. A.C. amps, 0.6 mA- 30 A, accuracy ± 1.5% F.S.D. ohms, 0-1 M, accu- racy ± 1.5% full ± 6% true value at mid-scale. The instrument is shock-proof and overload protected, measures 212 x 110 x 82 mm (8.5 x 4.4 x 3.3 inches) and weighs 1 kg (2.2 lb.). A range of accessories is available. One-off price is £55. New Varactor Diodes The ITT Components Group is introducing a number of new types of varactor diode. This VUE series consists of an improved range of step-recovery mesa diodes with a screened epitaxial structure. These components can be used in frequency multipliers with output frequencies from 4-8 GHz to 10-14 GHz. The VUE series comprises the tuning diodes VSA413H, VSA417H, VYA413H and VYA417H, all of which are suitable for the VHF and UHF bands as well as for microwave frequencies. The minimum quality factor at 50 MHz is 1000 for the 41 3H types and 800 for the 417H types. r ELECTROVALOE LTdI Telephone (061 1 432 4946. Shop hours: 0*ly 9-6.30 p.m.. 9-1 Sots U.S-A. CUSTOMERS »re inuitsd 10 conwct ELECTROVALUE AMERICA. Hannover Messe’Tr l S J I6.-Z4. April # l# Wir stellen aus: Elektor MOS Clock and iponc in stock We specialise in high-quality components at unbeatable prices. Full range of Elektor boards ex-stock. MOS Clock (two boards) £ 1-90 TV Sound £1-30 High quality Disc Preamp £ 0-85 Aerial Amplifier £ 0-85 ELECTRONICS 283 Edgware Road, London W2. Tel. 01-262 8614 ELEKTOR VERLAG GMBH D-5133 Gangelt 1, W-Germany Tel. (W. Germany) 02454-5055 Hours of business 9.30 - 6.00 Monday to Saturday Now you can change record-speeds without changing record-speeds. WeVe done away with the turntable and pick-up arm. old turntable speed-control, on The tracking error of the practically frictionless this very advanced Philips GA209 pick-up arm is very small, record deck. Side thrust compensation is adjustable for all Simply by placing a record on playing weights for both spherical and elliptical styli. the turntable the correct speed is The top cartridge from the Super M range, the electronically chosen and the GP412, is supplied as standard, pick-up lowered gently into the run-in groove. At the end of the record the turntable stops and the arm returns to the rest. This facility ensures that both the record and stylus are fully protected. In manual operation, the pick-up can be positioned over the grooves and lowered by means of a touch control. The mechanism permits very accurate positioning. Controlled by a servo motor via electronic touch controls, it can be operated whether the deck is used manually or as a fully automatic deck. Electronic control makes sure that the turntable speed is kept constant. Separate fine speed controls for 33 '/j and 45 rpm. allow the record to be tuned to the pitch of any musical instrument. The photo-electric stop switch is completely soundless and frictionless. High stability and insulation against shocks and vibration are ensured by the floating suspension of the PHILIPS Sim ply years ahead