up-t o-date electronics for lab and leisure EUSHTIir ID I Digital watch a single-chip with day and date Morse decoder converts morse into running script Master oscillator (2) two designs for use in organs February 1976 40p c call-sign generator : .«•- . -ado amateur knows that his (or her) call sign must be transmitted at least once per ten minutes during a GSO. This requirement can easily be fulfilled by using an automatic call generator which transmits the call >«n ton minutes in morse. 206 212 e decoder - R. ter Mijtelen ju'pment described in this article converts morse coded information i i and numerals), thus considerably simplifying the reception of morse o normal alphanumeric characters coming soon / irritated? / announcement to our subscribers 224 speech processor 225 morse typewriter — R. ter Mijtelen 226 "» use of morse is avoided like the plague by most radio amateurs. Once the R. A.E. has been passed the morse Lf* s often relegated to the junk box. This is a pity as morse telegraphy has advantages over telephony, in terms of bandwidth and for long-distance working. To encourage those with an aversion to 'key-bashing' the ■one typewriter was developed, which makes CW operation child's play. The missing link digital wrist watch *>*fereas a couple of years ago a digital watch was available only as a two chip design costing around £ 40 for *■* two 10's, it is now possible to obtain a single 1C that will perform all timekeeping functions, and will aaolay date as well as time, for around £ 10. Two watches are in fact described in this article. A low-cost, -eatmely easy to construct version, and a more miniature version for the advanced constructor. driving lessons a tale that needs some further explanation. When logic circuits are driven from 50 Hz mains pulses, unsatis- ‘actory behaviour can be expected since the exact pulse shape requirements for proper functioning of the »ates cannot, usually, be met by pulses derived straight from the sinusoidal mains. Th«s article discusses those requirements and establishes a few basic principles in connection with digital logic driving pulses. digital master oscillator (2) f-c owing last month's discussion of methods of producing the top octave of an electronic organ (or other uySoard instrument) using a single master oscillator, two practical designs for such an oscillator are presented simple fm test generator . market 255 212 — elaktor february 1976 4UTO/H4TIC C4LLSIGN GENERATOR According to the condition of authoriz- ation and the legal rules and regulations, a radio amateur must transmit his call letters at least three times at the begin- ning and end of each transmission. Dur- ing the transmissions, the call letters must be included at least once per ten minutes in the text. The manner in which this should be done is not further specified. This offers the possibility of using an automatic call generator, since there is no objection to transmitting the call sign in morse. It will, however, be clear that the transmitted signal must meet a number of requirements. Design considerations Besides the fact that the call sign must be repeated several times, the conditions of authorization also mention the permissible transmission types and the maximum permissible bandwidth. For ordinary telegraphy with an unmodu- lated carrier (Al) a maximum band- width of 0.2 kHz is specified. The band- width actually used depends on the way in which the carrier is switched. Since the carrier would normally be switched with a morse key, it is sufficient in this design to use a relay which replaces the key. The same relay can be applied for FI -modulation (frequency shift keying). It remains to be seen, however, whether it is possible to combine A 1 for the call sign with one of the A3 modulations (S.S.B.) for the speech and, similarly, whether FI morse can be combined with F3 (FM or PM) speech. At the transmitter either of these combinations will involve little difficulty. For the receiver, however, only a combination of Al morse and A3 A or A3J speech might give reasonable results, because then the BFO in the receiver is switched on. From the above considerations it fol- lows that Al and FI modulation, although normally considered ‘ideal’ morse, really do not qualify for matic transmission of the call sign in Every radio amateur knows that his (or her) call sign must be transmitted at least once per ten minutes during a QSO. This requirement can easily be fulfilled by using an automatic call generator which transmits the call every ten minutes in morse. The call generator described here can be oper- ated automatically and manually, so that if required a QSO can be started and ended with the call in morse. elektor february 1976 — 213 »«»<»*<*«»«*# course of an A3 or F3 speech trans- mission. Consequently, the relay output would seem redundant. It can, however, be of excellent service in tone- modulated telegraphy. The contacts of the relay can then be used for switching over from microphone amplifier to tone generator. The tone generator can be of a simple design; for instance a square-wave gener- ator followed by a selective filter. The filter is necessary to limit the band- width. To obtain the purest possible signal, an active filter is used in this de- sign, so that an almost sinusoidal signal is obtained. Furthermore the circuit is designed to give ‘soft’ on and off switching of the signals, so that the bandwidth of the signal remains well within the limit of 2.2 kHz (A2) or 3 kHz (F2). For SSB the bandwidth is of course no problem at all in this case. Another point for consideration is the character capacity and the programming method. The maximum number of call letters is six. This applies to practically every country. Since each letter con- tains a maximum of four dots and/or dashes - only the digit that usually forms part of the call is longer - a mem- ory capacity of 24 positions should be more than sufficient. If the circuit is also to be used in mobile transmitters, it must be possible to extend the memory capacity to accommodate the /A or /M suffix. A memory capacity of 32 positions will be sufficient in practi- cally all cases - the theoretical limit to the length of the call sign is 34 pos- itions. Considering that not all letters consist of four dots and/or dashes, this memory capacity will often be too great. The theoretical minimum is 10 positions, e.g. ET3TEE. This is, how- ever, no objection since the circuit can be so designed that the memory can be expanded or reduced as required. The overcapacity of the memory can, of course, also be used for adding a start ( ) and stop signal ( ); this will improve the legibility of the very short CW message. When determining the memory capacity it was assumed that the spaces between the dots and dashes need to be pro- grammed separately; after each dot or dash a space equal to one dot will auto- matically follow. Furthermore it is assumed that no separate place in the memory is needed for programming a letter space (length three dots). This letter space will, however, have to be programmed in some way. The programming of dots and dashes can be achieved in a simple way: the apparatus transmits a dot unless a dash is programmed. Since the call generator must always produce the same sequence, a ROM (read only memory) is the obvious choice. Considering that pro- gramming is quite simple, this ROM can be a diode matrix. It can be scanned by means of a shift register. For generating the dots and dashes the choice fell on two adjustable mono- 214 - elektof february 1976 mors-o-mat stable multivibrators, with a 1 :3 ratio of the timing capacitors so that they pro- duce pulses in the same time ratio. Monostable multivibrators are used be- cause they offer the possibility of de- signing a kind of programmed oscillator which steps through the sequence at the correct rate despite the absence of a clock generator. The final design consideration concerns the choice of ICs. Two IC families qualify. It would seem attractive to make use of TTL ICs because of their low price. However, they do involve quite a num- ber of drawbacks. The switching threshold is rather low (about 1 V), so that the sensitivity to interference is correspondingly high. Furthermore, the sensitivity to strong HF fields is also quite high. Another disadvantage in this particular application is the relatively high switching speed of TTL ICs, as this can easily give rise to spurious signals at very high frequencies. These drawbacks apply to a far lesser extent for the Cos-Mos family. The switching threshold is no less than 45% of the supply voltage. At a sufficiently high supply voltage the interference sensitivity can be one tenth of that for TTL ICs. Furthermore, the highest oper- ating frequency is lower, so that the switching speed is correspondingly less. Another great advantage of the Cos- Mos ICs is their low current consump- tion. Of the two IC families mentioned, the Cos-Mos family is undoubtedly the best choice for this design. The circuit The heart of the circuit consists of two monostable multivibrators, which deter- mine the length of dot and dash respect- ively. Both multivibrators are on one chip: the CD 4098 or the (MC 1) 4528 (see figure 1). Both types have the same pinning and show hardly any differences in specification. The prefixes CD and MCI indicate the manufacturer, other prefixes also occur. The pulse time of the multivibrators can be calculated fairly accurately with the formula T x = Rx ' C x , where R x and C x are the external time-determining elements. It is desirable to approach the dot/dash ratio of 1:3 as closely as possible. The high input impedance of the Cos-Mos IC means that a high resist- ance value can be used for the time- determining element. This has the ad- vantage that the capacitors can be kept relatively small, so that no electrolytic capacitors (with their high tolerances, typically -10 and +60%) need be used. The deviation from the dot/dash ratio depends mainly on the external com- ponents: since the two monostable multivibrators are mounted on one chip, the differences between them will be negligible. Even if the pulse times — and hence the transmitting speed — are variable, the ratio of 1:3 in pulse times must be maintained. This means that the ratio of the timing capacitors must always be 1:3, whilst the potentiometers which set the pulse widths must be coupled; a stereo potmeter is the obvious choice Since components with exactly the nominal value occur only in fairy tales, the ordinary commercial types will have to do. As regards the capacitors this means a tolerance of 5% and for the potentiometers a mistracking of less than 6 dB. In practice it is found that the ratio 1:3 can be reasonably well approached with these components, so they are obviously better than the speci- fications would lead one to expect! If necessary, the ratio of 1:3 can be very accurately approached by slightly altering one of the capacitor values (trimming). The multivibrators are followed by. simple selection logic (N8 . . . N15, see figure 2a). Depending on the program- * ming this logic will pass on a dot or a j dash. Feedback via N7 . . . N5 ensures , retriggering of the multivibrators at the end of each dot/dash or space. Simul-j taneously FF1 receives a clock pulsej from N16, so that this flipflop changes elektor february 1976 — 215 state. Consequently dot or dash and space are produced in turn as the relay is switched on and off via T1 . The same output of the flipflop is used to shift the memory (figure 2b) one position further. Since this shift register responds only to positive-going edges, this implies that it can change position only after two MMV times. At the end of the total sequence the start flipflop (N1/N2 fig- ure 2a) is reset via a differentiating net- work. The timer (555) connected as an astable multivibrator with an extreme pulse/ pause ratio triggers the start flipflop every five to ten minutes, after which the programmed call letters are auto- ! matically transmitted. A more detailed explanation of the cir- 1 cuit can be given by investigating how one total sequence is completed. , Operation The entire circuit consists of a closed I loop: the call generator starts and stops itself. Starting and stopping is controlled by the SR flipflop consisting of N1 and N2. The sequence can be started either auto- matically (by the timer) or manually. For manual operation SI applies one (or several!) negative edge(s) to the input of N 1 The output of N 1 then becomes logic ‘1* and the output of N2 becomes logic *0’. The same happens when the timer (IC3) produces its very short negative pulse. The positive part of the five to ten-minute period of the timer (this is the charge time of C2 to the triggering point) is determined by the sum of PI , R1 and R2. The formula for this delay time is: t = 0.693 (PI + R1 + R2)C2. The length of the negative starting pulse is determined by the dis- charge time of C2 via R2 to the second triggering point. The time for this part of the period is given by the formula: t = 0.693 R2 • C2. Considering the value of R2 the dis- charge time of C2 will be about 1 /8 sec- ond. This is sufficiently short to ensure a reliable cycling of the programme. are still ‘0’ (all shift registers reset), N13 and N 14 are still blocked. Furthermore FF1 is reset, so the Q output of this flipflop is also ‘O’. Consequently the oscillator (N17 + N18) and N10 are blocked. Thus the only signal that can be transmitted is a short space: the duration is determined by MMV2, and FF1 is reset. Because the output of N1 1 becomes logic ‘O’, a ‘1 ’-level will appear across R9. This level changes back to ‘0’ as soon as the period time of MMV2 has elapsed. The negative edge then occur- ring triggers the pulse shaper consisting of N7 and N6. The negative-going edge of this pulse triggers FF 1 via N 1 6 and the positive-going edge retriggers the MMVs via N5. The short delay between the end of the first MMV period and the start of the second has very little effect on the ratio of 1:3. The positive-going edge of the Q output of FF1, coinciding with the end of the MMV period, triggers the shift registers via line ‘A’. The shift registers are 4015s (see fig- ure 3). Figure 1. The heart of the circuit consists of two monostable multivibrators. The 1C 4089 (or 4528) comprises two MMVs on one chip, so that the characteristics of the two MMVs are equal to within very close tolerances. Figure 2. The complete diagram of the call generator. The abbreviation SP stands for space. The diode programming shown in fig- ure 2b is given as an example. Figure 3. The shift register is one of the most important parts of the memory. The type 4015 used here contains two four-bit As long as the SR-flipflop is in the ‘stop’ position, MMV1 and MMV2 are blocked by a logic ‘0’-level on the reset inputs. As soon as N1 becomes logic ‘1’, the MMVs can be triggered with a negative edge at their ‘TR’ inputs. To avoid triggering problems the ‘O’-level from N2 is converted into a short negative pulse by N3 and N4. N5 inverts this signal so that the negative trailing edge will trigger the MMVs. The outputs Q1 and Q2 of the MMVs now change to logic ‘1’. The selection logic determines what happens next. Since both programme lines (B and C) The IC4015 comprises two four-bit shift registers which must be triggered by a positive-going edge. The reset inputs are activated by a logic ‘1 ’-level. The shift registers are series-in/parallel- out types. A logic level on the data input is shifted to the A , -output on the positive-going edge of the clock signal, and from there on down the chain. Since, initially, all shift registers are reset, IC9, IC10 and 1C1 1 will only pass on ‘O’-information. The first shift regis- ter receives its data input from FF2. Initially, this flipflop is reset, so the 0 output is logic ‘1’. This is the infor- 216 - elektor february 1976 mation which is shifted to the A t -output on the positive-going edge of the first clock pulse. The positive-going edge then occurring at the A i -output of the shift register sets FF2 so that the infor- mation at the data input of IC8 goes from logic ‘1’ to logic ‘O’. As a result of the Q-output of FF1 becoming logic ‘1’ the oscillator (N17/ N18) is able to start. Furthermore the blocking of N10 and N13 is removed and Nil and N14 are blocked instead, so that now the programme line (C) de- termines what signal will be transmitted. Since the first signal to be transmitted in this example is a dash, the pro- gramme line becomes logic ‘1 ’ via the diode from output A 1 of IC8. N13 is now open and N10 remains blocked. A logic ‘l’-level occurs across R9, as be- fore, but the duration is now that of a dash. At the end of this dash the cycle described above is repeated. The MMVs are retriggered again and the state of the output of FF 1 changes. The oscillator is blocked and a space is transmitted. De- pending on the programming on the space programme line (B), the length of the space is equal to a dot or dash. In this example a short space is pro- grammed. The switching cycle is repeated until the last memory position necessary for the total sequence is reached. This memory position (in the example it is output 3 1 of the shift register — i.e. pin 1 1 of IC11) is connected to the reset line (E). As soon as this shift register output becomes logic ‘0’ again, the SR flip- flop N1/N2 will be reset. The circuit then waits for the next start pulse, after which the sequence is repeated. As appears from photo 1, the call let- ters PA 0 HKD given in this example are neatly reproduced by the prototype. The signal displayed on the photograph was taken from the Q-output of FF1. The audio output gives an entirely different picture (photo 2), The square- wave signal generated by the oscillator is first fed to a voltage divider which re- duces the signal amplitude to a level acceptable for the 741 (IC13). C9, R15 and R16 form a highpass filter. The opamp is used as an active filter which converts the square-wave signal into a quite acceptable sinewave. At the same time it ‘softens’ the switching of the signal. A low harmonic distortion of the sine and a ‘soft’ switch on and off of the signal are necessary to attain a narrow bandwidth. The results obtained with the given circuit are shown in photo 2. This clearly shows the gradual attack and decay characteristics of the signals. The dark lines in the center of the picture are due to residual distortion of the sine wave: remnants of the original square wave are seen as a kind of cross- over distortion. This distortion is a use- ful aid during initial trimming of the oscillator: when the oscillator is set to the correct frequency (with P3) the out- put amplitude is at its maximum and the cross-over is located at the zero crossing of the sine-wave. In the prototype the oscillator fre- Photo 1. Oscilloscope pulse diagram of th* call ( — PA 0 HKD — — ) programmed in the example given. This signal was taken from the Q output of FF1, so the upper half of the signal shows the programming. Resistors: R1 = 390 k R2 = 180 ft R3.R9.R12.R21 = 220 k R4 = 1 M R5,R6,R1 1 =470 k R7,R8,R10,R13 = 100 k R14 = 1 k R15.R16 = 150 k R17 = 68 k R18.R19 = 5k6 R20 = 2k7 PI = 470 k. lin. P2 = 2x 470 k, lin. P3 “ 1 M, preset Photo 2. The opening sign, as it appears at the audio output. The oscillogram shows the very gradual switching on and off of the dots and quency was 8 1 0 Hz with the component values shown and using standard com ponents. Small variations due to com ponent tolerances are possible, but thi frequency will never exceed 1 kHz so that the bandwidth of the signal fall amply within the requirements. The printed circuit board Figure 4 shows the p.c. board and com ponent layout. The board is designed for a memor with 32 positions. If necessary, th capacity can be extended or reduced a required Extensions will, however, hav to be mounted outside the p.c. board. PA 0 HKI 218 - elektor february 1976 morse decoder R. ter Mijtelen The equipment described in this article converts morse coded infor- mation into normal alphanumeric characters (letters and numerals), thus considerably simplifying the reception of morse messages. 1 The decoding equipment receives the message to be decoded in the form of an audio frequency morse signal at the out- put of an existing communications re- ceiver. Each incoming character is stored until it is complete and then decoded and displayed on a twelve-segment alpha- numeric display. A number of displays could be arranged in a line so that com- plete words could be displayed as run- ning script. The displays used in the prototype were home-made, but those who do not wish to construct their own displays could use a commercially- made display such as the Litronix Data Lit 16. The equipment comprises four sections. A signal processing section, which amplifies the audio output of the re- ceiver, rectifies it and shapes the pulses into TTL compatible logic levels; a shift register into which the processed signal is fed to assemble it into parallel form; a memory, and a decoder. The signal processing section (figure 1) operates as follows: the incoming audio signal is amplified by T1 and T2 and rectified by D1 and D2. Cl removes the A.C. component of the signal, leaving a series of D.C. pulses with some superim- posed a.f. The pulse train is amplified and limited by T3 and T4, and finally a Schmitt trigger S 1 produces a TTL com- patible output. The result, at the output of SI, is thus a TTL pulse train, the lengths of the pulses being equal to the duration of the tone bursts that made up the dots and dashes of the morse audio signal. The Shift Register Since the morse characters are trans- mitted serially (i.e. the dots and dashes that make up a character are transmitted one after the other) it is necessary to assemble each character in parallel form before it can be decoded. This is the function of the shift register of figure 2. The morse pulse train from output A of the signal processor is fed into the input of a serial in-parallel out shift register and is clocked through the register to appear in parallel form at the outputs. The maximum number of bits that the shift register must store is determined by the longest morse character, which is numeral ‘O’, consisting of five dashes. Starting with a dot as a basis (1 dot oc- cupies 1 bit in the shift register) then a dash, which has a duration of three dots will occupy 3 bits in the shift register, and the spaces between dots and dashes, being of one dot duration, will occupy 1 bit. Thus the longest character, zero, consisting of five dashes with four spaces between, requires a total of 19 bits. Thi is thus the required capacity of the shift register. The shift is made up of three cascade* 74164 8-bit shift registers, though not all the available bits are used. The circuit of figure 2 operates as fol- lows: As soon as a morse character is receive* then input A will go high on the first dot or dash of the character. Via NS thi sets the set/reset flip-flops N1/N2 an* N6/N7. The output of N6 takes thi clear inputs of the shift registers high s« that they can accept new data. The out put of N1 takes the reset inputs of 7490 1C1 low so that it can then coui clock pulses from the clock pulse genei at or S2. The 7490 is connected as divide-by-5 counter, so the pulses th appear at the output of N4 to drive th clock input of the shift register are i one-fifth of the clock frequency. The character is thus eventually a sembled in the shift register as a sequent of logic ‘IV and *0Y. A sequence * three IV corresponding to a dash, single 1’ corresponding to a dot and "0’ corresponding to a space betwe* dots and dashes. During the space when input A goes low the reset inpu of IC2 will go low and IC2 will corn Figure 1. Signal processing stage of the decoder. The output, point A is connected to point A of figure 2. Figure 2. Circuit of the shift register and memory. clock pulses from S2. However, while dots and dashes are still appearing at input A IC2 will be reset every time in- put A goes high. At the end of the morse character, however, there will be a space of three dots duration before the start of the next character, and this space is recognised as an indication that the character is complete and may be stored in the memory and decoded. When input A goes low at the end of a character IC2 can then count clock pulses unhindered. If no further data has appeared at in- put A by the time 1C2 counts the 10th clock pulse then the output of N3 will go low, resetting flip-flop N1/N2 and thus IC1 . With its reset inputs held high by N1 IC1 can no longer count, so no further pulses reach the clock input of the shift register. At this stage therefore, the morse character is held in the shift register, with a 2-bit space COY) on out- puts A and B of the left-hand 74164 due to the 10 clock pulses counted after the completion of the character. On the 1 1 th clock pulse the output of N8 will go low. The output of N10 will thus go high and the data present on the regis- ter outputs will be entered in the memory, which consists of 5 7475 4-bit latches. On the 12th clock pulse the output of N10 will go low again, dis- abling the clock inputs of the 7475’s so that the data is stored in the memory. Finally, on the 13th clock pulse the output of N9 will go low, resetting flip- flop N6/N7. This clears the shift register, in preparation for the next character. The morse character stored in the mem- ory is now decoded and displayed. The Decoder The latches in the 7475 have both Q and Q outputs, so both the data and its complement are stored. This is very use- ful as both the data bits and their comp- lements are used in the decoder. The decoder is split into three sections. The outputs of the memory are first decoded into ‘groups’ (figure 3a, b). The groups are then decoded into one of 36 outputs for each character (figure 3c, d) and fi- nally the 36 outputs are decoded into a 1 2 segment format to drive the display. Table 1 shows how each morse character appears at the outputs of the memory (complements not shown) with a cross representing a logic ‘1’. A group of three crosses, of course, represents a dash, a single cross a dot, and no cross, a space. It can be seen that there are groups of crosses common to several characters, where dots or dashes coincide in differ- elektor february 1976 - 221 Figure 3. The decoder circuitry. The AND- gates in figures 3a and 3b are the group de- coding. The group decoder outputs are further decoded in figures 3c and 3d to give 36 outputs, one for each letter of the alphabet and each numeral. ent characters. These groups are ident- ified by numbers corresponding to the outputs of figure 3a/b. For instance, numerals 1,2,3 and 0 all contain group 13. There are in all 38 groups. The 27 basic groups shown in table 1 are decoded by the circuit of figure 3a; a further 11 combinations are decoded according to figure 3b. The relevant groups for each character are then combined by the cir- cuit of figure 3c,d which gives 36 unique outputs, one for each character. Note that in figures 3c and 3d the circled outputs are the decoded alphanumeric outputs. Uncircled outputs are the group outputs (all outputs of figures 3a and 3b). Where an input is marked with an uncircled letter this corresponds to a direct connection to the appropriate output of the memory. An uncircled number means a connection to the out- put of the appropriate group decoder. It can be seen that some of the group decoder stages require only inputs from the memory, others require inputs from the outputs of other group decoder stages, and still others require a combi- nation of both. The same is true of the alphanumeric decoders. Decoder E 222 — elektor february 1 976 morse decoder V m \ ■ 7 []-U*~ Z-7*_ E-t-F P-l •=> 1- 1 :M,: F4 • \ 2-1*7 GO* n R=P* \ 3-2*- H-l 1— 5-5*' i H- l* L 1 -1 L4 •_ 5-5 GO LH l*_ 5-5*1 K-l/* \ I/-I •/ 7-“*/ R=H* L=l *_ W=l 1 * a B-H-I B-E- < M-l l* v X-y- X 3-5* 1 E-L* N-ll*\ y-/- x 0 (figure 3c), for example, requires inputs from group decoder 9 and also direct from output A of the memory. In figure 3a to 3d the numbers inside the gates refer to the type of gate required, thus 20 = 7420. The final stage in the decoding is to change the 36 outputs of the alpha- numeric decoder into 1 2 segment format . This is the function of the circuit of fig- ure 5 . Since the outputs of figure 3c and 3d are normally high (i.e. when an output is active it is low) figure S is de- signed to accept low inputs. Figure 4 shows the display font of the twelve- segment readout, and shows how certain characters are built up from simpler characters or segment groups. The outputs of the circuit of figure 5 are connected to the lamp driver circuit of figure 6, which drives the display. If a commercial LED display is used rather than a home-made display then of course segment resistors must be included in series with the output of the inverters to limit the current. Running Script As mentioned earlier it is possible to arrange several displays in a line to form running script that travels to the left along the line of displays. The circuitry to do this is given in figure 7 and oper- ates as follows. While a morse character is being received output B of figure 1 is 224 — alektof february 1976 morse decoder high, which means that the reset inputs of the 7490 in figure 7 are also high and the counter is in the reset state. On completion of the character output B goes low, and the 7490 can now count clock pulses provided by S3. On the first clock pulse output 1 of the 7442 will go low. The output of the inverter connected to it will go high, so that the data on the outputs of the three 7475’s labelled M3 will be transferred to the outputs of latches M4. Thus, momen- tarily the latches will have the same data on their outputs. On the second clock pulse output 2 of the 7442 goes low, so the data from M2 is transferred to M3 and so on. Finally the data on the decoder outputs a to 1 is transferred to the outputs of Ml. On the ninth clock pulse output 9 of the 7442 goes low, inhibiting the clock pulse generator until the counter is again reset by in- put B on receipt of a new character. The effect is that each time a new character is received the display shifts one place to the left. Although only five sets of latches and displays are shown the system can be extended to a maximum of 8 displays by connecting extra latches and dis- plays with the clock inputs of the latches driven from the unused outputs (5 to 8) of the 7442. Editorial Comment Although this system is ingenious it is felt that in practice great difficulty would be experienced in matching the clock rate of the equipment to the sig- nalling speed, especially when receiving manually sent signals. The system could, however, be used successfully with the ‘morse typewriter’ by the same author, since a signalling speed could then be agreed upon and adhered to quite easily. Such a system would have advantages over more conventional RTTY since the output of the typewriter could be received and taken down by hand, which is not possible with the output of a conventional teleprinter. H coming soon Front-end for TV sound this will convert the existing design (Elektor 2, p.236) into an indepen- dent receiver for tv (and fm) sound. Automatic rhythm generators add-on units for the minidrum that will give several rhythms at the flick of a switch. TV tennis extensions additions to the basic game (Elektor 7, p.l 1 1 1) giving new games, noises and scoring. Audio preamplifier a low-cost, high performance pre- amplifier and control amplifier with ‘remote control’ capability. SSB receiver a sensitive (0.5 pV for 12 dB S/N) receiver for single sideband trans- missions. DNL dynamic noise limiter. irritated? Are you irritated by this digital issue? In our introduction to Elektor (Elektor 1 , p.5) we warned you that this might happen! In this particular issue, admittedly, our hand was slightly forced: the ‘morse-designs’ belong together, and part 2 of the digital master oscillator follows part 1 in the previous issue. This does not leave much room for anything else. However, we do try to maintain a reasonable balance between the various types of circuits (simple or complex, digital or analogue). This means that those who are overjoyed with this issue may well be irritated next month, when we try to do without *0’s and ‘l’s. On the continent, where our readers have grown accustomed to a certain amount of unpredictability, we often hear the phrase ‘Wir lassen uns geme uberraschen!’ - which can be roughly translated as ‘Go on, surprise us!’. We intend to. announcement to our subscribers Dear Subscriber, Your subscription for Elektor volume 1 ran out with number 9. In order to adjust our administration for 1976 you will have received a renewal card by the end of December. The renewal-subscription rate for 1976 (February- December) is £ 5.80 to UK addresses and to all countries by surface mail. To all countries by air mail the subscription rate is £ 10.40. All prices are inclusive p&p. Please note that number 15/16 (July -August) is a double issue ‘Summer Circuits’, price 80p. If you wish to renew your subscription, you are kindly requested to sign this renewal card and to send it to your bank, which will take care of settlement. If you write a cheque please send it together with the lower half of the renewal card to our administration. Overseas subscribers who wish to convert their subscription from air mail to surface mail or vice versa are requested to pay the corresponding rate and at the same time to notify our administration. Thank you for your co-operation. elektor february 1976 — 225 speech processor To obtain the maximum efficiency when modulating a transmitter the modulation depth must be kept as high as possible for as much of the time as possible. This means that the modulating signal ampli- tude must be kept reasonably constant. Since speech most definitely does not have a constant amplitude some form of processing is called for. The most commonly used methods of speech processing are clipping (cutting off the signal peaks) and dynamic com- pression (reduction of the dynamic range of the signal to achieve a reason- ably constant signal level without dis- torting the waveform). The disadvantage of clipping is that it operates only on the amplitude peaks. It cannot boost a low level signal so that a low modu- lation depth may still occur. On the other hand, if the signal level is increased so that even low level signals give a reasonable modulation depth, then the peaks will be very severely clipped, resulting in distortion and loss of intelli- gibility. Dynamic compression effectively boosts low level signals and cuts high level signals, thus achieving a fairly constant mean signal level. Unfortunately, due to the relatively slow response time of dynamic compressors, transient peaks may not be effectively suppressed, and overmodulation may occur. The circuit described here overcomes these difficulties by combining both dynamic compression and clipping. The signal is first compressed to achieve a reasonably constant mean signal level, and is then clipped to remove any peaks. T1 and T2 form the microphone ampli- fier. The gain of this stage depends on the impedance of the microphone used, so that a high impedance, high output crystal microphone will produce the same sort of output levels as a low impedance, low output dynamic micro- phone. This avoids large variations in the level of the signal fed to T3 when using different types of microphone. R5, C5, D1 and D2 form a voltage controlled attenuator. A control voltage is fed back from the emitter of T4 to vary the forward bias voltage on Dl. If the base voltage of T4 exceeds the voltage at the anode of D3 by about 0.5 V then the signal fed to the base of T3 is attenuated by R5, C5 and Dl. R23 may be switched in or out to vary the compressor time constant. The compressed signal is taken from T3 via C8 and CIO. Any peaks are clipped by D6 and D7. The degree of clipping depends on the ratio R8 : R9. A low-pass filter is provided comprising T5, R17-R20 and C11-C14. The values giVen are suitable for operation in the 80 m band, where from 3 kHz upwards there should be a roll-off of at least 14dB/octave. For working in other bands where the filter is not required points A and B may be joined and the passive filter components omit- ted. If a different turnover frequency for the filter is required then the capaci- tance values Cl 1 - Cl 4 should be multi- plied by the factor |, where f is the required turnover frequency in kHz. Thus, for a frequency of 6 kHz, the capacitance values would need to be halved. As a final constructional point it must be stressed that diodes D 1 to D7 should be of reputable manufacture. Many of the ‘unmarked untested’ diodes on the market have a forward voltage drop of up to 1 V, and the circuit will not operate satisfactorily with these. When using the specified 1N4148, the voltage at the anode of D3 should be about 1.5 V to 1.7 V. u 226 — elektor february 1976 morse typewriter TYPE A morse signal, of course, consists of dots and dashes. The duration of dots, dashes, and the spaces between them bear fixed time relationships to one another. Starting with the dot as a basis, a dash has a duration of three dots. The space between dots and dashes has a duration of one dot, that between com- plete letters three dots, and the space between words has a duration equal to six dots. On this basis it is easy to work out a time scale for the duration of each letter of the alphabet, in morse. This is shown in table 1 . One cross represents a dot, and a row of three crosses, a dash. The longest duration letters are J,Q and Y, which have a duration of 13 dots, including spaces. For simplicity, punc- tation marks and numerals have been omitted from the basic design, but they may easily be added if required. The basis of the morse typewriter is a timebase generator that can produce 1 5 sequential outputs (plus rest position) corresponding to the 15 positions in table 1 . The timebase is used to drive an encoder to produce an output which is the morse code for the particular key depressed. The encoder is effectively a read only memory (ROM) in which are stored the bits of all the morse charac- ters. The ROM is addressed by the out- put of the keyboard and the output of the timebase generator to read out the required morse character. The circuit of the timebase generator is given in figure 1. It consists basically of a variable frequency clock pulse gen- erator based on a TTL Schmitt trigger, which drives a 7493 4-bit binary coun- ter. The binary outputs of the 7493 are decoded into one of 16 outputs by a 74154 decoder. When no key is de- pressed the reset input to the timebase 1 elektor february 1976 - 227 R. ter Mijtelen The use of morse is avoided like the plague by most radio amateurs. Once the R.A.E. has been passed the morse key is often relegated to the junk box. This is a pity as morse telegraphy has advantages over telephony, in terms of band- width and for long-distance working. To encourage those with an aversion to 'key -bashing' the morse typewriter was developed, which makes CW operation child's play. is high (this function will be explained in the discussion of keyboard operation). When a key is depressed the reset input goes low and the 7493 counts clock pulses. On completion of the morse character the counter is reset. Where in the count this occurs depends upon the length of the character. This is denoted by the letter R in table 1 . For instance, the shortest character, E, requires only a single dot, so the counter is reset on the second clock pulse, whereas the |N longest characters, J, Q and Y, have a duration of 13 dots, so the counter is reset on the 14th clock pulse. In order to produce letter spaces and word spaces it is necessary to inhibit the ^ keyboard for a duration of three dots or six dots respectively after the com- pletion of a character. This is the func- tion of the lower part of the circuit of * figure I . While the reset input is low and the 7493 is counting, the output of N1 is high, inhibiting further operation of the keyboard. The clear inputs of FF1 to FF3 are held low, so these flip-flops are reset. On completion of the character the reset input goes high. After this a letter Fipji* 1. Circuit of the timebase generator with space key. LED D1 indicates when the typewriter is ready for the next letter. Figure 2. Circuit of the group encoder, which produces groups of dots and dashes at various positions in the counter cycle. By combi- nation of the appropriate groups using NOR- gates all the letters may subsequently be 228 — elektor february 1976 morse typewriter ]7^)o — °* ’O ^N2^0 OB 6 0 |ni^Q — oo — ° c — oq :|=g)o — or | i> 60 )n7^Q °G Sec 20 [_)n2I^O OT > cc N1.N4.N5.N8 -7402 N9,N13,N14,N19 -7402 N20.N21 - '/] 7402 N2.N6.N7 - 7427 N11.N12.N18 -7427 30 ljr> — oi — ° v N18.N22.N23 - 7427 N24.N26 “ 7/3 7427 N3.N10 - 7425 ° J « 0 }n23^Q— - ow N25 = <67426 — ° K ,j|Eg>>-^> v