ISSN 0970-3993 Rs. 10.00 75 July 1989 n r< M THE MAGAZINE FOR PROFESSIONALS » Plucking the fruits of robot » In-circuit transistor tester * 8-digit frequency meter ► OtOS switches for audio applications » Semiconductor : C.R. Chandarana Address : ELEKTOR ELECTRONICS PVT. LTD. 52, C Proctor Road, Bombay-400 007 INDIA | Telex: (011)76661 ELEKIN OVERSEAS EDITIONS Route Natlonale; Le Seau; B.P. 5< 59270 Bailleul - France Editors: D R S Meyer: 5100 Aachen - West Ge Editor: E J A Krempelsa Elektor EPE Karalskakl 14 16673 Voula- Athens - Editor: E Xanthoulis Elektuur B.V. Ingelek S.A. Plaza RepCiblica Ecuador 2-28016 Madrid -Spain Editor: A M Ferrer Electronic Press AB Box 5505 14106 Huddinge- Sweden Printed at : Trupti Offset Bombay - 400 0 Copyright ® 1989 Elektuur B.V. Volume-7, Number-7 July-1989 CONTENTS Editorial The mission must succeed 7.07 Special Features Plucking the fruits of robot 7.1 1 Application notes Voice recorder from Texas Instruments .. 7.43 Components CMOS switches for audio applications 7 -22 Practical filter design (6) 732 Semiconductor diodes 7 - 3 ^ Electrophonics PROJECT: MIDI keyboi .. 7.35 .. 7.54 MIDI Signal redistribution 757 Computers PROJECT: In line RS 232 monitor 7.48 Test Measurement PROJECT: 8-digit frequency meter 7.26 PROJECT: In-circuit transistor tester 7.46 Information Industrial licence and letters of intent 7 59 Electronic News Industry News Telecom News ... Guide Lines Corrections Classified ads Index of advertisers . 7.20 ,. 7 17 7.15 ... 7.73 ... 7.74 ... 7.74 7.05 Front cover Perhaps the major feature of the MIDI- compatible keyboard controller published in this issue is that it can be used with practi- cally any existing keyboard, whether salvaged from a discarded instrument, or still in use in a piano, organ, or non- Mror synthesizer. It sup- ports up to 96 keys covering 8 octaves. THE MISSION MUST SUCCEED Commissions and committees have been rather too many in the country that the recent constitution of the Telecommunication Commission is likely to be dismissed as one of the so many commissions and nothing more. The country can ill-afford such a cynicism. Or at least, the Telecommunication Commission cannot be bunched with the powerless and purposeless governmental organisations. That a person like Mr. Sam Pitroda heads the commission and that the commission has been formed not a day sooner than required compels us to look at this commission with a difference. Elsewhere in this issue, we have dealt with the most basic issues like why telephone services cost too much in this country and why don't we have sufficient number of telephone systems which work efficiently. Often, we notice a widespread tendency to blame the telephone department for all the owes, though it cannot be totally absolved of the blame. Outdated and overused equipment and cables are certainly the cause of the poor phone services . Most of all, the traditional attitude that telephone is a luxury and cannot be accorded priority in the planning process ensured a primitive slot for telecommunications in India. Undoubtedly, this attitude has changed and there Is an awareness that telecommunication services are inseparably linked to the economy and progress of the nation. Even If one looks at telecommunications purely in a commercial angle, still it deserves a special treatment. Telecom services and equipment In India are worth more than Rs. 4000 crores. International trade is a crucial factor nurturing the health of a country and international telecommunication is a part of it. India, like any other nation simply cannot remain isolated in the globalisation of communication. It may be reasonable to assume that politicians, policy makers and planners have come to realise the potential of telecommunication and that they do not need much more prodding. As an expert has sounded a caution, the requirements of telecommunication in India are somewhat unique and variegated. The needs vary according to the segment like international trade, national business, civic life, rural areas and so on. Since these segments represent distinctly different markets, the strategies to be adopted should also be such as to suit each segment. There can be no single, uniform prescription to the ailments of the country in telecommunication. 7.07 CMOS SWITCHES FOR AUDIO APPLICATIONS T. Giffard When about ten years ago the first analogue CMOS switches and multiplexers reached the audio components market, many audio enthusiasts believed that there was at last an end in sight to the use of expensive relays and other electromechanical elements to control volume and rumble or switch signal sources and functions. Unfortunately, the low speeds, high, non-linear on-resistance and level of crosstalk associated with the new devices soon put an end to these expectations. Over the past few years their quality is claimed to have improved considerably. These claims have been tested in our laboratory through a number of CMOS switches and circuits. We will commence by taking from the numerous parameters of CMOS switches those that are of importance to audio designers, namely: ■ resistance of the closed switch (Ron in Q); ■ analogue voltage range (U« in V); ■ Ron as a function of U« (in %); ■ consistency of Ron over a number of switches (in %); ■ insulation in off condition (in dB); ■ crosstalk between a closed and an open switch (Ct in dB); ■ rise time (Ton in ns); ■ drop-out time (Totr in ns). The first four of these parameters are particularly important for the linearity of the audio circuit; the next two, for the crosstalk performance, and the rise time is of vital importance in some appli- cations as we shall see later. Fig. 1. This simple circuit is perfectly satisfactory for many audio applications. Topology of CMOS switches CMOS switches may be used for three specific functions: (1) the selection of the signal source; (2) switching of auxiliary functions, such as changing filter characteristics or altering the volume, in the same way as a rotary switch; and (3) as quasi-digital volume control. In (1) and (2) the basic circuit of the switch is almost always the same: it serves to interrupt the signal path in a fairly simple manner. For instance, in Fig. 3. An improved version of Fig. 1 for Fig. 5. Typical crosstalk vs frequency charac- Fig. 6. An improved version of Fig. 3 for t.ie most demanding applications. 7.22 Fig. 7. Schematic diagram of a DC-controlled preamplifier. Fig. 8. Traditional high-quality electronic volume control covering a range of % dB in 2 dB steps. Fig. 1 the popular CD4066 has been in- serted into the signal path to serve as a relay or electromechanical switch. The 10 kS load resistance is part of a general audio network. Tests of this circuit were reasonably satisfactory in spite of the dependence of Ron on the signal level and supply voltage (typical curves of the former are given in Fig. 2). The rela- tively large value of Ron and that of the ratio Ron:Ri caused some distortion of the signal. The tests also showed that CMOS switches, even from the same manufac- turer, vary quite a lot from one to another. The overall distortion varied from -74 dB to -84 dB (<0.02%), depend- ing on the IC, at a supply voltage of ±7.5 V and a signal level of 1 V r.m.s. The distortion remained within the values indicated when the signal level was increased, but increased sharply when the supply voltage was reduced. This IC can not be recommended for use in exacting applications, but for normal purposes it is perfectly satisfactory. The fact that the non-linear drop across the switch at high signal levels was the cause of much of the distortion led us to the circuit in Fig. 3. This has a much better distortion figure: -87 dB (0.0045%) at a supply voltage of ±5 V. When the supply voltage was increased to ±7.5 V, the distortion could no longer be measured accurately. This would mean that this circuit is suitable for even the most exacting audio re- Fig. 9. Alternative to Fig. 8 with electronic step control via single CMOS switches. quirements, were it not for the channel elektor India July 1989 7.23 separation and crosstalk (-84 dB at 1 kHz; -60 dB at 20 kHz). Although the measured figures would be satisfac- tory for mass-produced equipment, they are not for good-quality apparatus. TVpical characteristics of these parameters are given in Fig. 4 and Fig. 5. It may also be considered a drawback of the circuit that the opamp inverts the signal. A further improvement of the circuit is shown in Fig. 6. This has an additional CMOS switch that short-circuits the signal when the switch in the signal path is open. The control signals for the two switches must therefore be in antiphase. The circuit shows an improvement in crosstalk and channel separation to - 84 dB at 20 kHz. At this frequency the layout of the PCB makes a greater con- tribution to the distortion, as we have found many times in the design of audio equipment. 7.24 elektor India july 1989 it diagrams of the CMOS switches in Table In the choice of a type of switch, quality, available space on the PCB, and price, play a role. If quality is deemed the most important factor, it is best to use single- switch ICs. On the other hand, if price is important, there are analogue multiplex- ers that contain a number of switches in one housing (just like stepping switches). These ICs save money and space. How- ever, as will be seen from Table 2, a num- ber of parameters of these devices are considerably worse than those of single- switch devices. CMOS preamplifier The circuit of Fig. 6 may be used to form an important part of a complete preamplifier, a basic design of which is shown in Fig. 7. The source selector may be a 2xl-form-8 multiplexer. The volume control may consist of two 1- from-8 multiplexers as shown in Fig. 8, or of single CMOS switches as shown in Fig. 9. If auxiliary functions, for in- stance, bass lift or stand by, are required, they may be realized with the aid of single CMOS switches. The control logic is also fairly simple to design as shown in Fig. 10. This circuit is based on two start-stop oscillators, Ni and Ns respectively. NAND gates Ns and N« generate the appropriate signal for 6-bit counter ICs-IO. At the same time, the state of monostable N2-N4 determines whether IC3 will count up or down. The outputs of the counter may be connected direct to the volume con- trol in Fig. 8. Make sure that the level of the control signal to the logic circuits and of that to the CMOS switches are the same. Volume control by signal ratio An interesting application of fast CMOS switches is shown in Fig 11. The four switches are clocked by astable multivibrator N1-N6 at a frequency of 100-150 kHz (sampling theory holds that the clock frequency must be at least twice as high as the highest audio fre- quency). Switches Si (S4) and S2 (Sa) are provided with control voltages that are in an- tiphase and are, therefore, never open or closed at the same time. The duty cycle is determined by the setting of Pi. The 'lumps’ of audio signal at the out- put of the switches are fed to ICi. This opamp serves as a low-pass filter — - (for removing the clock signal); as an in- tegrator (for synthesizing the lumps of audio signal; and as an impedance con- verter. The circuit as shown receives two audio signals whose attenuation is inversely proportional to their loudness: the louder channel A, the softer channel B. Many variations may be applied to the circuit without affecting the original audio signal: one channel may be omit- ted; Pi may be replaced by the circuits in Fig. 8 and Fig. 10; and others that we will leave to the reader’s ingenuity. 7.25 8-DIGIT FREQUENCY METER by T. Giffard A state-of-the-art frequency meter module is presented that has an 8-digit, 7-segment LED indication, a resolution of 10 Hz, and accepts input frequencies of up to 3.5 MHz. Its presetting facility makes this simple-to-build module ideal for incorporation in a radio receiver. 10 000 000 - 0 107 000 = 09 893 000. shift right (10 Hz); MSD borrow; preset = 99 989 300 The counter module has an up/down input and a separate, but optional, circuit for programming the offset. Resolution The module is based on two ICM7217IPL CMOS presettable up/down counters. Two of these chips are cascaded to obtain an 8- digit read-out on common- anode 7-segment LED displays. The counter's presetting facility makes it eminently suitable for use as a fre- quency read-out in recei- vers, since the intermediate frequency (e.g., 455 kHz or 9 MHz, can be programmed as an offset. In this manner, the output frequency of the local oscillator (L.O.) may be measured by the counter module, when driven by a suitable prescaler. Depend- ing on whether the L.O. fre- quency is lower or higher than the received frequency, the IF offset is divided by the prescale ratio and then pro- grammed as a preset value, which is automatically added to, or subtracted from, the module's input frequency to ensure that the received frequency is shown on the display. An example might help to illustrate the above procedure. A super-heterodyne VHF FM broadcast receiver has an inter- mediate frequency of 10.7 MHz. The L.O. frequency is higher than the received fre- quency. Assuming that the receiver is tuned to a station at 100.0 MHz, the L.O. generates 110.7 MHz. This signal is ap- plied to a divide-by-100 prescaler, which drives the frequency meter module. To ensure that the display reads 100 MHz, the counter must be programmed for an IF offset of 10.7 MHz/100=107 kHz. Since the counter will normally count up, it must be set to a negative offset, the one's- complement of this frequency, which is simple to calculate as and gating times are simple to change, if desired. The maximum input frequency of the counter module is about 3.5 MHz at a sensitiv- ity of 60 mVrms. The counter chip The ICM7217IPI is a CMOS decade counter in a 28-pin plastic enclosure, intended for being programmed with the aid of switches or fixed logic configurations, and driving common-anode dis- plays. The device from GE- Intersil (second source; Maxim) is one of a family of single-chip 4-bit pro- grammable up/down counters with an on-chip multiplex scan oscillator for simple driving of 7-segment LED displays. The internal structure of the 1CM7217 is given in Fig. 1. Three main outputs are provided; CARRY/BOR- row for cascading with fur- ther 4-bit counters, ZERO which indicates when counter state zero (0000) is reached, and equal which indicates when the current counter state equals the value loaded into the internal register via the BCD I/O pins. The three outputs and the BCD port are TTL-compatible and inter- nally multiplexed. Output CARRY/BORROW goes high when the counter is clocked from 9999 to 0000 when counting up (input U/D logic high), or from 0000 to 9999 when counting down (input U/D logic low). The Schmitt-trigger at the COUNT input provides hysteresis to pre- vent double clocking on slow rising edges. The counter contents are transferred to the multiplexed 7-segment and BCD out- puts when input STORE is pulled low. A low level at the reset input causes the counter to be asynchronously reset to 0000. As already noted, the BCD port can function as an input or an output. These functions are selected with the logic levels applied to the three-level LOAD counter (ld) and load register (lr) inputs. When both are open, the BCD port provides the multiplexed BCD display selection sig- nals, scanning from MSD (most-signifi- cant display) to LSD (least-significant display). When either lr or LC is taken high, the BCD port is turned into a 4-bit input for loading the counter (lc) or reg- ister (lr) data. Since the ICM7217IPI is designed to drive common-anode dis- plays, the levels applied to, or provided by, the BCD port are 'high true'. When input lr is made low, the BCD I/O lines are switched to the high-imped- ance state, and the digit and segment dri- vers are turned off. The counting operation continues, however, and the re- maining input and output functions oper- ate normally. The displays are normally switched off with the aid of input lr to reduce power consumption during stand- by conditions. The on-board multiplex scan oscillator controls the internal timing of the ICM7217. The nominal oscillation fre- quency of 2.5 kHz may be reduced by con- necting a capacitor between input scan and the positive supply line. The oscilla- tor output signal has a relatively low duty factor to delay the digit driver outputs and thus prevent 'ghosting' effects on the displays. The digit and segment drivers on board the 1CM7217 are capable of directly driving common-anode 7-segment LED displays at a peak segment current of 40 mA. At a duty factor of 0.25, this corre- sponds to 1 0 m A per segment. Finally, the display CONTROL input rec- ognizes 3 logic levels. When it is logic high, the display segments are inhibited. When it is logic low, the leading zero blanking feature is turned off. Displays on with leading zero suppression is achieved by leaving the input open. Practical circuit As shown in the circuit diagram of Fig. 2, a pair of ICM7217IPIs is used in conjunc- j tion with a central timing generator type ICM7207IPD (ICi). This chip controls the gating of the input signal with the aid of 1 an external quartz crystal, Xi, inverter Ti and input amplifier T2. In addition , the 1CM72 071PD provides the STORE and RESET signal for the counter chips, IC2 and IC3. Although the store output of the ICM7207IPD is of the open-drain type, and the associated inputs of the ICM7217S have 75 pA pull-up resistors, an external pull-up resistor R2, is fitted to ensure im- munity to noise. The u/D and reset inputs also have internal pull-up resistors, and may, therefore, be left open for normal operation as an up-counter. The block di- . agram of the ICM7207 is given in Fig. 4. Monostable IC4 enables the counter to load the preset word. The LOAD COUNTER pulse is delayed with respect to the RESET pulse because the counter can only b e prese t with data other than 0000 when RESET is inactive. The preset frequency is set with two 1 blocks of 4-way DIP switch blocks. The circuit diagrams of these (optional) units are given in Fig. 3. BCD thumbwheel swit- ches may be used as a more ergonomical alternative to the DIP switches. Alterna- tively, wire links may be used if the counter works with one, fixed, preset fre- quency. The BCD port lines and the scanning digit selection signals are available on Ki and K3 for connecting to the preset unit. A few suggestions are given for those who want to experiment with the circuit. The duration of the count window may be reduced from 100 ms to 10 ms by tying pin 11 of the ICM7107IPD (RANGE con- trol) to the positive supply line. This modification results in a corresponding reduction of the counter's resolution, however: with pin 11 at +5 V, this is 100 Hz instead of 10 Hz. In both cases, a good-quality 6.5536 MHz quartz crystal is required: for optimum stability of the read-out, a type with 10 ppm tolerance or better is recommended (most inexpensive computer crystals do not meet this speci- fication). For high-resolution applications, the duration of the count window may be in- creased by a factor 10 (100 ms or 1 s) by using a ICM7207A in combination with a 5.24288 MHz quartz crystal. Unfortunate- ly, this is not a standard frequency, so that this crystal will have to be made to order. Pin 23 of both counter chips is con- nected to ground, so that leading-zero suppression is not used. As already dis- cussed, this feature may be useful in a number of applications. Where it is re- quired, pin 23 of ICs may be left open to achieve leading-zero suppression on the most-significant display group. Leading- zero suppression of the full 8-digit display may be realized by driving the display CONTROL input pin of the LS group driver, IC2, with the collector signal of a n-p-n transistor whose base is driven by the ZERO output of the MS group driver, IC3. In a number of cases, it may be possible to omit the two MS displays altogether. Resistor R17 is only required when the module is used without a prescaler. De- pending on whether a MHz or kHz indi- cation is required, the resistor lights the decimal point on LDs (MHz read-out) or LD2 (kHz read-out). Three receiver mode indicators, D33, D34 and D35, are provided on the display board. The LEDs may be controlled from the mode selection switch in the receiver. Three boards: a compact frequency read-out The lay-out of the printed-circuit board for the universal counter is given in Fig. 5. The PCB is cut into three to separate the preset unit (at the top), the main counter board (at the centre), and the read-out section (at the bottom). The receiver mode indication board forms a separate unit, which need, however, not be cut from the display board. Populating the boards is straightfor- ward and requires hardly any comment. It is strongly recommended to use sockets for all integrated circuits, displays and DIP switches. K2' and K4' on the display board, and K2 and K4 on the main counter board, are 16-way IC sockets with turned pins. These receive 16-way IDC pin- headers fitted at the ends of an approxi- mately 5 cm long flat-ribbon cable. The 10-way connections between the main counter board and the read-out are made in 10-way flat-ribbon cables. Pin-headers Ki and IO on the main counter board are fitted at the component side, and KT and K3’ on the preset board at the track side'. The pin-headers are con- nected with IDC sockets pressed on to the ends of an approximately 5 cm long flat- ribbon cable. Fig. 4. Internal structure of the ICM7207A timing generator (courtesy GE-Intersil ). 7.29 Parts list Resistors (±5%): Ri=10M R 2 ;Ra;R&-Ri 6 incl.=1 OK R4=1K0 Rs=27K Re=470R R7=1K5 Re=12K Ri7=33R (see text) Rts=220R Capacitors: Ci;C2=100p C3»80p trimmer C4=1 Op; 25 V; tantalum Cs-InO Ce=27n C7=4700p; 10 V Ca;C9;Cio=100n Semiconductors: Di-D32 incl.»1N4148 D33;D34;D35= LED ICt-ICM7207IPD (QE-Intersil or Maxim) IC2;IC3=ICM7217IPI or ICM7217IPJ (GE-ln- tersil or Maxim) IC4-4538 j LDi-LDe incl.-MAN72A (General Instru- ment Optoelectronics) Ti;T2-BF494B Miscellaneous: K2;K2';K4;K4'- 16-way DIL socket with mat- ing I DC plug. Ki;Ki’;K3;K3'- 10-way pin header with mat- ing IDC socket. Si-Ss incl.— 4-way DIL switch block. Xi- 6.5536 MHz quartz crystal. PCB Type 880128 L The construction of the flat-ribbon ca- bles that interconnect the sub-modules is illustrated in Fig. 6. Contrary to what some retailers of specialist tools would have you believe, IDC (.insulation displace- ment) connectors are simple to fit on to flat-ribbon cable with the aid of a carefully operated vice, or even a small hammer and two pieces of wood. Insert the cable between the socket or plug and the asso- ciated plastic cap, and align the individual wires with the clip-type connectors. Then close the connector by carefully pressing the cap on to body of the connector. Alter- natively, carefully tap the cap in place with the aid of a small hammer. Check the continuity at all pins. The completed sub-assemblies are then ready for mounting together in a sand- wich construction. The read-out board is mounted on top of the main counter board with the aid of three 25 mm long spacers or lengths of M3 threading. Make sure that the soldering connections of the re- ceiver mode LEDs, and those for the near- by terminal posts, do not touch the body of the large electrolytic capacitor, C7, underneath. The preset board is fitted 16 -way I I — IDC »i | ‘ Dip header I I U- 16 - way flatcable 10 - way I IDC socket 0= 10 - way llatcable =B Fig. 6. Construction of the lour flat-ribbon cables that interconnect the sandwiched boards. back-to-back below the main counter board with the aid of 20 mm long PCB spacers with internal threading. The com- pleted three-board assembly is shown in the introductory photograph of this ar- ticle. The unit may be installed in a receiver and connected to a regulated and well-de- coupled 5 V power supply. In some cases, it may be necessary to screen the module to prevent interference in the receiver. The readability of the displays may be im- proved by fitting them behind a red bezel. Calibration is simple if a frequency meter is available: adjust trimmer C3 for 6.5536 MHz measured at pin 5 of the ICM7207. Alternatively, tune the receiver for zero-beat against a frequency refer- ence station, and adjust the trimmer until the correct received frequency is dis- played. Sensitivity of the prototype was 35 mV™,, over 200 kHz to 1 MHz, and 60 mVrms at an input frequency of 3 MHz, Average current consumption with eight displays on (indication: 8x'8'), but the re- ceiver mode LEDs off, was measured at approximately 450 mA. Offset programming Assuming that the counter operates in the UP mode, and that the local oscillator fre- quency is higher than the received fre- quency, the required preset value is first converted to its 8-digit one's complement. Next, the corresponding DIP switches are set until the preset appears on the dis- plays. Examples for 455 KHz, 900 kHz (9 MHz with +10 prescaler) and 107 kHz (10.7 MHz with +100 prescaler) are given in Table 1. Always remember that the counter can not handle input frequencies higher than 3.5 MHz, so that the effective- ly programmed offset is the IF frequency divided by the prescale factor. For most SW and general coverage receivers, a +10 prescaler is suitable; for VHF receivers a +100 prescaler. PRACTICAL FILTER DESIGN - PART 6 by H. Baggott In this sixth part in the series we start our discourse of the tables and characteristics of filters and as first we deal with those pertaining to the Butterworth type because that is the best known and probably also the most often used kind of filter. The Butterworth filter owes its popu- larity to a combination of flat amplitude response in the pass band and reasonable roll-off. A drawback is its non-linear phase characteristic. The roll-off is fairly precisely 6 n dB per octave, where n is the order of the fil- The Butterworth filter may be consid- ered a compromise between the Bessel network (moderate roll-off but linear phase response) and the Chebishev filter (steep roll-off, poor phase response and ripple in the pass band). For applications that require a flat pass band and steep roll- off, the Butterworth filter is undoubtedly the best choice. Table 1 gives the pole locations of Butterworth filters of the second to the tenth order. These data enable the ready computation of filters with the aid of for- mulas given in earlier parts in this series. Butferworfh tables The dimensioning of filters becomes much simpler with the aid of Tables 2 to 5,which give component values for passive and active filters of the second to the tenth order. The values given always refer to a filter with a cut-off frequency of 1 Hz. Table 2 gives component values for a passive filter with identical source and output impedances. The component idcnti- cations at the top of the table correspond to those in the diagrams above the table and those at the bottom of the table corre- spond to the diagrams below the table. Table 3 gives the component values for a passive filter with negligible source impedance. Tables 4 and 5 give the component val- ues for active filters with a single feedback path. Table 4 deals with second- and third- order sections. If, for instance, you want to design a seventh-order filter, you take two second-order and one third-order section and connect them in tandem. It is also possible, as we have seen in Part 3, to use only second-order sections and, in the case of odd-order filters, add a passive rc network. The data for this are shown in Table 5. This table is given merely to illustrate the alternative way. Since in the majority of cases it is simpler to work with Table 4, Table 5 will not be given for the other filter types in future parts in this series. Butterworth characteristics For clarity's sake, the characteristics given in this article deviate slightly from those given as examples in Part 2. For each type of filter we will give three series of char- acteristics, showing respectively: the gain vs frequency response — Fig.32; the delay vs frequency response — Fig. 33; and the step vs time response — Fig. 34. The phase response is not given because this would not divulge all that much on a logarithmic scale. In any case, the phase linearity is easily deduced from Fig. 33, since lineari- ty corresponds to a constant delay time at 0.83147 0.55557 0.19509 7.32 ele Table 3. Normalized component values for passive low-pass sections with negligible source Impedance. o ™ o o — ^ — I O .» Cl C2 c. « C3 if ; **» Table. 4. Normalized component values for active filters with single feed- back path. Table 5. Normalized component values for filters with single feed- back path. Example 1. Design a passive low-pass Butter-worth filter with a cut- off frequency of 1600 kHz and a source and output impedance of 50£2. The attenuation at 3200 kHz must be at least 20 dB. Solution First we determine the value of the attenuation at each fre- quency relative to the normal- ized frequency of 1 Hz by dividing the reference fre- quency by the cut-off fre- quency: 3200: 1600 = 2. From Fig. 32 we determine which curve affords at least 20 dB attenuation at fi= 2 Hz. and this is found to be for a fourth- order filter the diagram of which is shown in Fig. 35a. Note that a third-order filter just would not do since it would give an attenuation of only 1 8 dB per octave. It would also have been pos- sible to deduce the filter from the diagram underneath Table 2. Study this carefully, be- cause once you understand this, the purpose of Table 2 will be clear forever. All that remains to be done now is to calculate the compo- nent values for the given input and output impedance and the cut-off frequency: C = C/(fR) L' = LR/f The calculations will be found to result in the compo- nent values given in the dia- grams in Fig. 35b; ' Similarly, the values for the components in Fig. 4a are found to be: all frequencies. Each of the figures gives the characteristics for a second-, fourth-, sixth-, eighth- and tenth-order section. Those for odd-order filters are assessed from intermediate values: this keeps the number of characteristics to a reasonable level to prevent loss of clarity. Note that in Fig. 32 for a clear view of the behaviour of the filter just below the cut-off frequency, the scale of the y-axis to the left of 1 Hz has been expanded and is shown at the left of the drawing. The val- ues of the gain at frequencies above 1 Hz are shown to the right of the drawing. Two examples We shall give a couple of worked out examples for each type of filter we deal with to give you the opportunity of learn- ing to use the tables and characteristics quickly and properly. Cj =0.1218/(1600000 x 50) = = 1.52 xl0' 9 = 1.52 nF L x = 0.2941 (50/1600000) = = 9.19x 10"® = 9.19 pH Example 2. Design an active fifth-order low-pass But- terworth filter with a cut-off frequency of illy 19B9 7.33 Fig. 32. Gain vs frequency characteristics of a Butterworth filter. Fig. 33. Delay time vs frequency characteristics of a Butterworth filter. 5 kHz. Solution. This is designed fairly quickly. It is an odd-order filter, so we need a second-order section and a third-order section, as drawn above Table 4. The two sections are connected in tandem, after which the normalized component values read from the table are inserted. Next, choose a value for the resistors (/? in the formulas), say, 4.7 kft. Then calculate with the aid of-the formula given in the first example (for C) the ‘real’ values of the components. Again, two examples of the calculations: C x = 0.515 / (5000 x 4700) = 21.9 x 10' 5 * * * 9 = 21.9 nF C 2 = 0.04918 / (5000 x 4700 = 2.09 x 10' 9 = 2.09 nF This completes our discourse on Butterworth filters. Part 7 will deal with Bessel networks. Fig. 36. Illustrating the computation of an active 5th-order filter. by D. Doepfer The feature par excellence of the MIDI-compatible keyboard controller described in this article is its ability to be used with practically any existing keyboard, whether salvaged from a discarded musical instrument, or still in function in a piano, organ, or non-MIDI synthesizer. Soon after the publication of the Port- able MIDI keyboard (Ref. 1), numerous readers asked us to give further details on the use of the Type E510 MIDI controller in conjunction with full-size keyboards of five and more octaves. This month we meet these requests with the description of a universal MIDI controller board, once again based on the E510, intended for use with many types of musical keyboard. The maximum number of keys sup- ported by the present design is no fewer than 96, covering 8 octaves. The controller provides the velocity parameter, and sup- ports one-octave transposition as well as instantaneous split-point programming to achieve data distribution between MIDI channels 1 and 2, with any key on the keyboard. The printed-circuit boards have been designed such that they may be used in conjunction with a keyboard hav- ing wooden keys and spring- or gold-wire contacts (Kimber-Allen type). Any other type of key or contact is, however, also suitable. A MIDI keyboard is classified as acces- sory equipment, not as an instrument, be- cause it is not capable of producing musical sounds. As such, it is used for controlling MIDI synthesizers ( expanders ), or micro-processor based systems run- ning special MIDI programs. The application range of the present UNIVERSAL MIDI KEYBOARD • universal polyphonic MIDI keyboard with a maximum of 8 octaves (96 keys) • transmits velocity parameter • 1 instantaneously programmable split- point (channels 1 and 2) • ±1 -octave transposition • simple-to-buikf circuit • circuit boards designed for use with spring or wire contacts • modular keyboard configuration allowed within maximum range of 96 keys: easy implementation of, e.g.. 54- or 72-key • inputs suitable for driving from contacts other than those on a musical keyboard • keyboard matched to controller either by software (EPROM contents) or hardware (physical connection of contacts) circuit is widened further by the fact that the key inputs are suitable for driving from almost anything that represents an electrical contact. We have, therefore, no reservations about calling the circuit universally applicable. To mention a few less usual, but technically interesting, ap- plications: key signals generated by the player interrupting light-beams, or actua- tion by weight of touch-sensitive areas on a theatre or dance floor. The velocity parameter is not always required for such applications, and is fair- ly simple to omit as will be shown later. Other ways of providing the key signals may come to your mind at this stage. At the end of the article, we describe an ex- perimental percussion interface to rouse your interest in finding new applications for the MIDI controller. We feel sure that the design will please many of our readers, who, nodoubt, will have their own follow-up suggestions for, say, a semitone transposition circuit, a sustain pedal, and typical MIDI functions such as program change, pitch bend and access to all 16 available channels. Let us know of such thoughts and ideas and we will respond appropriately. This two-part article describes the operation, construction and use of the universal MIDI keyboard. Although space did not permit a reiteration of the intro- duction to the MIDI keyboard, a descrip- tion of its principles and functions may be found in Ref. 1. This also discusses the way in which a MIDI keyboard controller circuit measures the time between the in- stant the pole of the key leaves its rest position and the instant it reaches the- 7.35 work contact. The present keyboard works on the same basis. 'ED EJ Fig. 1. Pinning and block diagram of the single-chip MIDI keyboard controller Type E510. Strike the right note with the E510 The Type E510 MIDI controller is without doubt a revolutionary integrated circuit, and has been recognized as such by many readers following the publication of the Portable MIDI keyboard. The plastic pack- age with only 16 pins (Fig. 1) contains a programmed control circuit with MIDI keyboard functions normally carried out by a fast microprocessor and one or more peripheral circuits. However, the E510 also has its drawbacks and limitations: it recognizes only one split, while up to 16 can be programmed on many keyboards. Also, the E510 can send data to MIDI channels 1 and 2 only. The velocity par- ameter can not be geared precisely to the characteristics of the keyboard, or be given the optimum range to suit the aver- age strike force of the user. Contrary to the single-chip, mask-pro- grammed E510, most microprocessor sys- tems are 'open' which means that they may be programmed or re-programmed to include the above features. The E510, on the other hand, has the advantage of being extremely simple to use in a practical cir- cuit. Acknowledging the fact that the vast majority of musicians working with MIDI equipment are not electronics buffs, a simple circuit is a significant factor. A number of readers have expressed their doubts and reservations about the dynamic range of the E510. These doubts are really not justified. In fact, the velocity processor in the E510 is so good that the chip is capable of distinguishing between a soft, normal and hard keystroke even when Digitast keys are used as on the Portable MIDI keyboard (Ref. 1). Digitast keys have tactile feedback which makes them quite unsuitable for providing vel- ocity information, as is clearly explained in the relevant article (this is not to say that the Portable MIDI keyboard is touch- sensitive in the sense specified by the MIDI standard). The present MIDI key- board is fully equipped for velocity pro- cessing, however, and the fact that it also uses the E510 is proof of our confidence in the chip. Before study ing the circuit and the con- tents of the transposition EPROM, get the right orientation by briefly looking at Fig. 2, the block diagram of the MIDI key- board. Constructors of the Portable MIDI keyboard will easily recognize the general structure. Circuit description To avoid an unnecessary large and clut- tered circuit diagram, Fig. 3 shows the (entirely theoretical) configuration of the MIDI controller with 16 keys only. The circuit diagram in fact shows only one of the possible six key decoders that may be installed. As a result of this simplification, the diagram is hardly any more complex than that of the Portable MIDI keyboard. As shown by Fig. 3, each of the six key decoders is capable of addressing up to 16 key contacts, so that a maximum of 96 key contacts is available (the grand piano keyboard has 88 keys). The circuit diagram of the keyboard section in two possible versions is given in Fig. 6 (its operation will be discussed in due course). As already stated, the basic operation of the E510 keyboard controller in the present application is similar to that in the Portable MIDI keyboard. Details of the key scanning mode and velocity processing are, therefore, not repeated here since these have been covered at length in Ref. 1. The E510 has an on-board 7-bit binary counter, which provides states 0 through 127 on outputs AO through A6. Between these outputs and the key contacts sits an Fig. 2. Block diagram of the universal MIDI keyboard controller. 7.36 elektor indl. July 1989 address transcoder in the form of an EPROM. This chip has two functions: first, it suppresses the E510-generated ad- dresses corresponding to notes so low that they are inaudible, and, second, it allows the player to select up or down transposi- tion of a section (zone) of the keyboard. The binary values that appear at the counter outputs of the E510 are applied to the address inputs of the EPROM. The output word of the EPROM is available on 7 data lines. Of the 7 output bits, 4 carry the address of one of sixteen keys within a decoded group, and 3 the address of one of six decoders. The actual key addresses are carefully programmed values to ob- tain either the normal mode with no split points, or up/down transposition of the counter values supplied by the E510. The 4 least-significant data lines (LS nibble) of the EPROM are connected direct to the binary inputs of l-of-16 decoders Type 74HCT154, which, in turn, are connected to the key contacts. The most-significant data lines of the EPROM (MS nibble) drive the address decoder, a l-of-8 decoder Type 74HCT138, whose outputs enable the six key decoders. With the exception of the 74HCT138, the keyboard interface is basically the same as that used in the Portable MIDI keyboard. The addition of the l-of-8 decoder and some modifications to the EPROM con- tents make it possible to increase the num- ber of keys to that required for a full-size MIDI keyboard. The relation between the keyboard type and the EPROM contents will be reverted to. known as split-point programming. Split-point Briefly, a split-point, or simply split, on a MIDI keyboard effectively splits the key- board into two smaller keyboards, whose size in terms of keys is defined by the player. The principle is illustrated in Fig. 4. On a 6-octave keyboard, for in- stance, the 2 low octaves may be assigned to a bass instrument on MIDI channel 1, while the higher 4 octaves are assigned to another instrument, say, piano accom- paniment, controlled via MIDI channel 2. The top part of Fig. 3 shows the split- point programming circuit. The E510 scans the keyboard in low-to-high order, i.e., from the key producing the lowest note to the one producing the highest note. A split is simply programmed by actuating push-button & simultaneously with the key that defines the wanted posi- tion of the split. This action causes the address of the key to be stored in memory. The output of the split-programming cir- cuit pulls input CO of the E510 high while the chip scans the keyboard, and a key is addressed with a number higher than that of the key that defined the split-point. The E510 responds to the high level at CO by redirecting all MIDI data to output chan- nel 2 rather to than channel 1. When the key scanner has reached the highest key, i.e., when the E510 has passed counter state 127, the split-programming circuit is reset, and co is made logic low again, so Fig. 5. Experimental percussion Interface (see Ref. 2). that MIDI data is routed to channel 1 The split-programming circuit can only store a key address when line BS is low, which is the case when the pole of the addressed key reaches the work contact, and S2 is closed. In that condition, gates N9, N4 and Ns generate a positive pulse transition at the CLK input of IC7. This octal bistable then copies the logic combi- nation applied to its inputs, D0-D7, to its outputs, Q0-Q7. The combination forms the address of the key actuated by the player programming the split. Bit D7 does not form part of this address: it is forced logic high and causes Di to light, indicat- ing that a split has been programmed. During subsequent keyboard scan cycles, 1C6, an 8-bit comparator, compares the address stored in memory and applied to its inputs B0-B6 to that available on the address bus of the E510 and applied to its inputs A0-A6. When these addresses are equal, i.e., when the keyboard scanner reaches the key that defined the split, the bistable formed by Ni and N2 is set to logic 1 by output A=B of the 74HCT688 (pin 9 of Ni). Input CO of ICi goes logic high. At the end of the keyboard scan cycle, the bistable is reset to logic 0 by the negative pulse transition on address line A6, which drives differentiator C3-R7-D3. When input CO of the E510 is low, MIDI data is routed to channel 1. When CO is high, it is routed to MIDI channel 2. At power-on, the bistable is reset to 0 by R7 - Cj. Octal latch Type 74HCT273 is also reset at power-on with the aid of a low pulse generated by Rs-Ct and applied to the rst input. Actuation of S2 when no key is pressed (BS is logic 1), causes network C4-R10 connected to N» to reset the latch also, while any previously programmed split is erased. Diode D2 protects the input of N2 against voltage peaks. In practice, it is. recommended to al- ways erase an old split before programm- ing a new one simply by pressing S2 only. It is possible to direct the 'low' key- board section to the left of the split to Fig. 6a. Configuration of an integral 96-key keyboard. Databyte 00 is loaded in the EPROM at relative address 12io, or OCh counting from the start of block 180 h ad- dresses in normal mode without transposi- tion. A keyboard with 72 keys starting with note Fmay ‘start’ on the second contact of the second lowest decoder (selected with link B). Non-used contacts may be left open, or connected to the BE line to simu- late the presence of rest contacts. In that case, the first decoder board, normally en- abled by link A, need not be installed. When it is desired to have, for example, 3 complete C-to-C octaves to the left of the middle C, the keyboard must start one oc- tave lower at the F note corresponding to MIDI KEY 17. In that case, the board se- lected by link A must be installed, while the last board enabled by link F may be omitted. A 54-key C-to-B keyboard, for instance, starts at contact S9 of the second board. 7.39 channel 2, and the section to the right of the split to channel 1, instead of the other way around which forms the default con- figuration. Two possibilities exist for this modification: • insert non-used inverter Nio (ICs) in the CO line (pin 12) of the E510; • break the connection between input co and the output of Ni (pin 8 of ICj). Con- nect input co to the output of N2 (pin 1 1 of IO) instead. This modification causes an 'unsplit' keyboard to address MIDI channel 2 instead of 1 at all times. Inverter Nio in ICs is useful when the velocity parameter is to be omitted. In that case, the rest contacts of the keys need not be connected because only the work con- tacts are used. Indeed, the keys need not have a rest contact at all. Line BE must, however, be forced high by the actuated BS signal, and be forced low when BS is inactive. To free the be input, remove pull- up Ri, and connect it to the output of Nio, whose input is connected to BS. This modi- fication is illustrated in Fig. 9. Percussion enthusiasts are referred to Fig. 5, which shows an interface that allows the keyboard inputs to be driven by signals obtained from a simple beat detector built from a piezoceramic buzzer (Ref. 2). Transposition by EPROM The first task of the EPROM is to place the physical keyboard in the range of 128 vir- tual keys addressed by the E510. The con- troller counts from 0 to 127 irrespective of the actual number of keys connected. Without a decoder or transposition cir- cuit, the lowest key on the keyboard vyould correspond to key MIDI 0. This is riot vey useful because this key number belongs to a subsonic frequency. The EPROM thus allows the real keyboard to be centred around number 60 of the 128 virtual keys. This centre is formed by the middle C as illustrated in Figs. 6a and 6b.. Since enough space is left in the EPROM, the complete physical keyboard can be transposed towards the low or high end of the virtual keyboard. This is the second function of the EPROM, whose available memory capacity is, however, still not exhausted. Therefore, jumpers Ji and J2 are provided to give access to nor- mally unused memory in the EPROM for the implementation of special functions. The jumpers are normally installed so that effectively the lower quarter of the Fig. 5b. Configuration of a 72-key key- board. The EPROM is re-programmed such that the first contact of the first decoder board corresponds to the first key of the keyboard. In normal (non-transposed) mode, databyte 00 (see Table 1) is loaded in the EPROM at relative address IDh, or 29io, counting from the start of block 0180h in Table 3 (this will be given in Part 2). 7.40 elektor india july 1989 Switch Si sets the logic levels on address lines A7 and A8. operation, up-transposition or down-transposition: ind so selects between normal j EPROM contents tor virtual keyboard with 96 notes fror *C = middle C on the first additional line under the treble st EPROM is used. Removal of one or both jumpers causes another, differently pro- grammed, address area to be selected in the EPROM. Details on programming are given in the relevant section below. Jumper Ji selects one of six enable sig- nals A-F in the key decoding circuit. There are 16 contacts to each keyboard sub-circuit. The number of decoders re- quired depends on the number of keys available on your keyboard. Jumpers are, therefore, placed to individual require- ment. Examples: a 4-octave keyboard re- quires at least 3 address decoders, a 64-key type 4, and a 72-key type 4VS as illustrated in Fig. 5. At least five decoders are required for 80 keys, 5 VS for 88 keys, and, finally, all 6 for 96 keys. The jumper for the first decoder (at the 'low' side of the keyboard) is marked A, the next one B, and so on, up to jumper F, which enables the decoder that reads the highest 16 keys. The standard EPROM contents corre- spond to a 96-key keyboard with a tone range from C (MIDI KEY NUMBER 12) to B (MIDI KEY NUMBER 107). Figure 6b il- lustrates the fitting of a 72-key keyboard with range F to E into the 96-key range addressed by the EPROM. The actual number of keys matters very little, pro- vided double addressing is avoided. More importantly, however, the number of the lowest key of the keyboard used must correspond to the counter value reserved for it by the E510. In other words, if, for example, a 54-key C-to-F keyboard is available, an EPROM may be used with the contents given in Table 1, but only if the lower C of this keyboard is connected to contact S9 of the second decoder board, as shown in Fig. 5. Modifying the EPROM contents to suit individual requirements is not necessary in most cases, but fairly simple on the basis of the information given below. Programming the EPROM The standard contents of the EPROM for a 96- key keyboard are listed in Table 1 . To facilitate altering the contents. Table 2 gives the unprogrammed 'framework' which serves to document one's own EPROM contents. Table 2 can be com- pleted by entering the actual key numbers as shown in the example of Table 3 (this will be included in next month's instal- ment). Having studied the circuit diagram of the MIDI controller, it will have been no- ticed that output bit D4 is not used. Nor- mally, bit 7 is not used, but here the design of the printed circuit board has forced the omission of bit 4. The upshot is that the most-significant nibble in the da- tabyte is always nought or an even num- ber (0, 2, 4, 6, or 8), as shown in Tables 1 and 3. Mind this simple rule when compil- ing and programming your own EPROM I with the aid of Table 2. Possible misgivings about the versat- ility of the MIDI keyboard should be dis- pelled by the fact that the EPROM may 7.41 hold up to 64 different keyboard configu- rations. Jumpers Ji and J 2 allow the selec- tion of 16 different tables. The remaining 48 are available after modifying the con- nections of address lines All and A12. Electronics enthusiasts not interested in electrophonics may like to know that the E510, in conjunction with a microproces- sor, is also eminently suitable for building an advanced multi-point contact scanner. References: 1. Portable MIDI keyboard. Elektor Electro- nics November 1988. 2. Disco drum. Elektor Electronics June 1984. The construction of the MIDI keyboard will be discussed in next month's second and last instalment of this article. Table 2. To program the EPROM: 1. Enter ‘O’ in the cell corresponding to the number of the lowest key on your key- 2. enter the successive key numbers in ascending order, right up to the highest key. 7.42 elektor India iuly 1989 DOWN-TRANSPOSITION (A7=1 ; A8=0) "C = middle C on the first additional line under the treble stave. APPLICATION NOTES Voice recorder from Texas Instruments For many years now, the most popular means of analogue recording and playing back of audio signals has been the cas- sette recorder. But even here, digital techniques are beginning to make inroads. True, available material allows only relatively short recording times, but for a number of applications, for instance, telephone answering machines, advertising messages, memory aids, alarm installa- tions, and so on, it is perfectly usable. A new ic from Texas Instru- ments, the TMS 3477. is intended as basis for such equipment. Apart from RAM, all necessary functions are available on the chip. The block diagram of a possible system is shown in Fig. 1 . The ic may be operated in two different ways. The simpler is by means of a four-position keyboard, of which the keys assume the functions corresponding to those normally available on a cassette recorder. The other method is via a computer. Dynamic rams instead of cassette tapes are used as recording medium. If you want to listen to something different, you insert a different bank of drams or make a new recording. A modified form of continuously vari- able slope delta modulation (CVSD) is used in the TMS 3477 for the quantization (digi- tization) of the audio signals. This type of modulation used with drams has the important advantage of requiring only simple connections between the TMS3477 and the drams. The principle of cvsd is shown in Fig. 2. The analogue signal, u x , is compared with Uy, a signal that increases or dimin- ishes only slowly. Whether u y increases or diminishes depends on u x , which in its turn depends on the difference between u x and Uy. The digital signal u x thus contains information on the analogue signal. Since it is a digital signal, it may be stored in a Fig. 1. A recorder system based on the tms 3477 and its block diagram. memory. Another advantage of delta modulation is that the integrator of the modulator may be used also as demodulator. Signal u y then serves as the output signal. The integrator (which is indispensable for delta modulation) is built up in the tms 3477 rather differently from what you might expect. It is constructed from an adder and a digital-to-analogue converter. The adder is the real integrator, since, in this case, integrating is nothing more than increasing the preceding result by 1 (if u x is high) or reducing it by 1 (if u x is low). The converter has been added to translate the digital con- tent of the adder into an analogue signal, Uy, which is either fed to (he comparator or, during play- back, to the output. Several of these stages may be recognized immediately in Fig. 3. First, there are the comparator, the data latch, the adder and the digital-to-analogue converter that form the delta modulator. To these are added two further inte- grators to enable the speed with which u y can change is matched to the signal level. This greatly improves the quality of the The remainder of the chip con- sists of the necessary control logic for the external memories and the host interface via which the tms 3477 is controlled. An experimental circuit diagram for a complete recorder system is shown in Fig. 6. The tms 3477 contains a mode register that defines the execution mode. This reg- ister is programmed at the power-on reset via the address outputs of the drams (AP0-AP9,where ap stands for Address/ Program), which serve as temporary input during the reset procedure. Since the-AP pins serve as inputs and outputs, the logic levels for initializing the ic MUST be applied via pull-down resistors ( R I- R 1 0 ) - pull-up resistors have already been provided on board the chip. Table 1 summarizes the functions that may be realized via these pins. The type of ram that will serve as memory for recording is set via pins apO and apI. There is a choice of 3: tms 4164 (64 Kbit); tms 4256 (256 Kbit); and tms 4cl024 (1 Mbit). Up to two rams (only of the same type) may be connected. Whether one or two are used is indicated via ap2. Switches si and s2 further extend the possibilities of the rams. When S2 is open, it is possible to select either of the two rams by Si. This enables two different phrases to be selected — PH(rase)l and PH(rase)2. With s2 closed and si in posi- tion ph 1 (obligatory), it is possible to record and playback one phrase which may, however, be twice as long as either PHl orPH2. The next setting refers to the length of playback period. This may be given a fixed value equal to the maximum, of which more later. With variable playback period, (too long) intervals at the end of a recording may be prevented. If after a Fig. 3. Internal structure of the tms 3477. " m m . (T) X S3477/3477A) 3 MIC AMP c, — — Q ^TO SPKR-JACK PWR E AMP , rJX T* PHRA^ ^ W 0 SELECT Ljj-I (TOGGLE SW) Fig. 7. Line change switches Table. 1. The tms 3477 contains a mode register that defines the execution mode. This register is programmed at the power-on reset via input pins ap0-ap9. These pins are also used as out- puts to address the external drams. The type of external drams used is programmed via these pins like the mode of interfacing the chip with either a keyboard or a microprocessor. This table is used for memory and interface selection and defining the type of use of the chip. recording the stop key is pressed, the memory address in which the last sample is stored is retained and this serves as stop address during playback later. Another method is cyclic recording, which is set by ap 4. With this method, the TMS 3477 continues recording until the stop key is pressed. Since with that method the memory will be full after a certain time, the new data is written over the old. The beginning and the end of the recording are thus ‘floating around’ the memory as it were. The memory therefore always contains the last section of the recorded audio signal, which is useful in, say, a dictating machine. The type of interface via which the tms 3477 is controlled is selected by ap 5. If the keyboard is selected, the voice recor- der becomes a manually controlled stand- alone unit. In this application, four switch- es are connected to the four interface in- puts. The function of these speaks for it- self. Controlling the tms 3477 via the CPU interface offers a number of possibilities, since the CPU allows the realization of a variety of ancillary functions, such as data transmission between two voice recorders or the storing of data in a large memory with the possibility of calling up several messages on command. Control is effected via those pins of the ic that are also used for the keyboard inter- face. The functions of those pins are total- ly different then, however. There are two Command Port lines (CPO and CPl), a data strobe (STB) and a busy signal (see Fig. 4). ‘ A high level on the strobe line indi- cates that a new command must be execut- ed. Which demand is indicated by cpO and CPl. The busy signal enables the processor to check whether the tms 3477 is in opera- tion to prevent any unneessary breaks in recording or playback. The sampling frequency is set via pins ap 6 and ap 7. Depending on the desired quality of the sound one of three available frequencies may be selected. The duration of playback may be cal- culated from the sampling frequency and the memory capacity and is memory capacity / sampling frequency From this relationship it follows that the minimum playback time is 1 second (64 Kb; 64 kHz) and the maximum playback time is 131 s (2 Mb; 16 kHz). A facility afforded by the digital inte- grator is data compression. This, in spite of its name, is a form of expansion of the audio signal. In this mode, bits are multi- plied by 4 (that is, shifted to the left by two bits) before they are applied to the digital-to-analogue converter. In this way, soft recordings are reproduced much loud- er, albeit with a resolution of only 8 bits. This mode can not be used when record- ing, therefore, because this would cause a severe deterioration of the sound quality. The last function, recording monitor, is set via pin ap 9. It enables listening in dur- ing the recording. Finally, it should be noted that the tms 3477 is not housed in the usual dil pack- age, but in one with a much smaller grid (0.070" = 1.78 mm) Source: The “TMS 3477 solid-state voice recorder” by Philippe Clement • Texas Instruments. IN-CIRCUIT TRANSISTOR TESTER by A. Rigby In electronic troubleshooting a transistor is generally not above suspicion until it responds correctly to the usual diode-tests with an ohmmeter. Before these simple test can be performed, however, the transistor must be removed from the circuit. Experience teaches us that this operation is time-consuming as well as possibly harmful to the PCB and the rest of the circuit in a good many cases, while it offers no guarantee that the cause of the malfunction will be found. The super-simple and inexpensive good/faulty indicator described here tests almost any transistor in circuit. A further useful feature of the tester is its built-in npn/pnp indication. The circuit shown in Fig. 1 is straight- forward and based on low-cost compo- nents. The central part is a dual J-K master/slave bistable Type 4027, ICi, of which one section, ICi», is configured as a multivibrator. The frequency of the symmetrical output signal is set to about 100 Hz by R1-R2-C1-C:. This signal is applied direct to the input of the second bistable, ICis, which supplies the tran- sistor under test (TUT) with two com- plementary-phase signals, Q and Q, which have a frequency of 50 Hz. In the absence of a TUT, current limiter Rs passes a current through one of the LEDs, Ds or Di. These are connected in anti-parallel and light alternately be- cause of the complementary drive sig- nals supplied by the bistable. Because the LEDs are turned on and off at a rate of 50 Hz, they appear to light virtually constantly to the human eye. Bistable outputs Q and Q are connected to a potential divider, R3-R4. The voltage at junction R3-R4, t/ii/2, is applied to the base of the TUT. A correctly functioning npn TUT con- nected to test terminals B, C and E is switched_on via D3 and D4 when Q is high and Q low, since the base is positive with respect to the emitter. Both LEDs then remain off: Ds because it is effec- tively short-circuited (the drop across an intact collector-emitter junction is about 0.1 V), and D7 because it is reverse- biased in that condition. When the bi- stable toggles, however, the transistor is turned off, so that Ds is reverse-biased, and D7 lights. The situation is reversed if a correctly functioning pnp TUT is connected: Ds then lights while D7 re- mains off. Spotting defective transistors Defective transistors typically have either a short-circuited or a broken col- lector-emitter junction. In the first case neither diode lights because of the con- tinuous short across them. A broken c-e junction gives the same visual indication as the absence of a TUT: the LEDs light alternately. Diodes D1-D4 are included to prevent the tester giving an ‘OK' indication with a transistor that has a base-to-collector or base-to-emitter short. This leaves only one semiconductor junction jn the tran- sistor, which then acts as a diode. Depending on the logic state of the bi- stable, either D1-D2 or D3-D4 drop about 1.2 V, which is added to the drop across the collector-emitter junction of the TUT. A correctly functioning and con- ducting TUT has a typical c-e drop of about 0.1 V. Added to the 1.2 V intro- duced by the conducting pair of diodes, this voltage is not high enough to cause the turning on of the (red) LED that should remain off when the transistor is switched on. Therefore, only one LED lights: the indication is ‘OK’. This changes, however, if the TUT has either of the above short-circuited junctions, since then the c-e drop becomes 0.6 V rather than 0.1 V. The resulting total drop of about 1.8 V (1. 2+0.6 V) across the LEDs causes these to light simulta- neously: the indication is ‘faulty'. Summarizing the above, transistors that are good are marked by only one LED (pnp or npn) lighting. All other indica- tions (both LEDs on or off simultaneous- ly) point to a faulty device. Construction The small printed-circuit board designed for the transistor tester is populated per the Parts List and the overlay shown in Fig. 2. The completed board is then in- stalled in a plastic case with battery com- partment. The tester is connected to the TUT with three flying wires with mini- ature, coloured and sleeved, crocodile clips. The ‘on’ push-button, npn/pnp in- dicator LEDs and, optionally, a transis- tor test socket, are mounted on to the front panel. Fig. 2. True-size track layout and compo- nent mounting plan of the printed-circuit board for the transistor tester. Parts list Resistors (±5%): Ri;R2=220K R3=220R R4=330R R5=270R Capacitors: Ci;C2=470n Semiconductors: Di...D6incl.=1N4148 D7 ;Db= red LED; dia. 3 mm ICi=4027 Miscellaneous: Si= push-to-make button SPST. Bti= 9 V PP3 battery. PCB Type 896029 7.47 IN-LINE RS-232 MONITOR by A. Rigby Serial links between computers and peripheral equipment based on the RS-232 standard are notoriously difficult to get going for the first time. Much of the frustration computer users suffer while connecting-up serial equipment is caused by their inability 'to see what is going on' on the data and handshaking lines. The small in-line signal monitor discussed here largely solves this awkward problem for almost any equipment sporting an RS-232 input or output. Connections, computer ports and cables claimed to comply with the RS-232 standard are so common these days that the original application of this serial in- terface is often forgotten or not even known. In computer land, it is a gener- ally accepted fact that virtually all ’non- standard' RS-232 links — even those of the so-called ‘zero-modem’ type — take a lot of valuable time to get operational. Not surprisingly, it is often desired to have a simple tool available for monitor- ing the activity of data and handshaking signals. Before describing the operation and construction of such a tool, it may be useful to give a brief recapitulation of the basic operation of the RS-232 inter- face itself. Standard RS-232: OK as far as it goes The signals available on a RS-232 con- nector, whether male or female, 9-pin or 25-pin, are in principle intended only to ensure correct transmission and recep- tion of data from so-called DTE (data terminal equipment) to DCE (data com- munication equipment). A DTE is gener- ally any data source, but it is usually a computer. A DCE is any device that con- verts data in a manner that allows this to be actually carried over some distance to a receiving system. The best known example of DCE is the telephone modem (modulator! demodulator). The RS-232 interface is specified such that DTE is linked to DCE by wires con- nected to pins with the same numbers on the connectors at both sides of the cable: DTE pin I goes to DCE pin 1 , DTE pin 2 to DCE pin 2. etc. (see Fig. 1). Similar- ly, the signal functions are assigned such that data transmission is optimum on this multi-wire, but essentially simple-to- make. cable (see Table 1). DTE-to-DTE = zero-modem All was well with the RS-232 interface until, in the early seventies, someone decided to transfer files between two computers (DTE) by hooking up their RS-232 outlets. Such a connection be- tween two DTE-type devices was not foreseen or, for that matter, specified or supported by the RS-232 standard, and obviates a good many handshaking sig- nals. The so-called ‘zero-modem’ shown in Fig. 2 is known by now to virtually any PC user as a simple 6-wire cable (excluding ground which is not, strictly speaking carried over a wire) with one interconnection, 6 — 8, on each connec- tor. In fact, the zero-modem is not a modem at all (whence its name): it mere- ly acts as a single DCE ‘seen’ by .both computers (DTE). The other, even simpler, solution to DTE — DTE communication is the two- wire link, also shown in Fig. 2. Since this provides only handshaking to each individual computer, and not between the two of them, it may cause problems at relatively high data speeds. For most PCs and compatibles running the simple COPY COM1: instruction, the troubles typically start at 9,600 bits/s. The attempts of some PC users to intro- duce handshaking for computer-to-com- Fig. 1. Basic wiring dagram of a standard DTE— DCE 25-way RS-232 cable. Two -wire DTE -DCE link puter file transfer often rely on smart but essentially non-standard usage of the serial ports. Hence, these experiments are machine-specific and do not, in many cases, guarantee satisfactory re- sults in other system configurations. Bearing in mind that the RS-232 stand- ard is still perfectly all right for every- thing it was originally designed for (bidirectional communication between DTE and DCE), it is fair to argue that a good deal of the compatibility problems experienced these days are caused by non-standard configurations and appli- cations. By now, however, we seem to have accepted that the rapidly expanding use of computer-based communication has caused non-standard applications of the RS-232 interface to outnumber standard applications by far. So far, in fact, that the RS-232 interface is often unjustly critized for needless complex- ity while used in configurations it was never designed to handle. Examples of RS-232-based, but definite- ly manufacturer-specific, serial inter- faces include those on PC-ATs (the famous 9-pin connector), on Postscript laser printers that can ‘talk back’ to the computer, on equipment sending a non- symmetrical line voltage (down to simple digital drive with +5 V), and on a host of dot-matrix printers, intelligent modems, scanners and other digitizers, all commonly used in the PC environ- ment. Time, therefore, for a simple tool that enables the ‘communication expert' to quickly locate a problem if the serial link is no great shakes. Circuit description The circuit diagram of Fig. 3 shows that the signal indicator is built with a num- ber of bi-colour LEDs, associated series resistors, two connectors, and a printed- circuit board to the design shown in Fig. 4. The tracks take all 25 pins of female 25-way D-connector Ki at one side of the board direct to the male D- conneclor, K 2 , at the other. Seven lines between Ki and K 2 are ‘tapped’ to drive bi-colour LEDs that indicate the current logic level. The seven signals thus moni- tored are generally considered indis- pensable for correct data transfer via most RS-232 links. As to the definition of the logic levels used on RS-232 datalines, remember that a logic one corresponds to a nega- tive voltage, and a logic zero to a posi- tive voltage (this does apply to the control and clock lines). Construction The printed-circuit board is small to en- sure that the RS-232 monitor is a handy Fig. 2. Some commonly used RS-232 connections. 'ig. 3. Circuit diagram of the in-line RS-232 monitor. LEDs are used to indicate the status of the main signals carried via the serial link. 7.49 Table 1. Fig. 4. Double-sided printed-circuit board for the RS-232 monitor. Pin Signal Function DTE DCE 2 12 13 14 15 16 17 18 19 20 21 22 23 24 25 CG TxD RxD RTS CTS DSR SDCD SCTS STxD RxC SEL TCK BSY chassis ground transmitted data out received data in request to send out clear to send in data set ready in signal ground data carrier detect in positive test voltage negative test voltage not assigned secondary DCD in secondary CTS in secondary TxD out transmit clock (DCE) in secondary RxD tn receive clock in not assigned secondary RTS out data terminal ready out signal quality detect in ring indicator in speed selector DTE in speed selector DCE out data line busy in and rugged test device. The copper is- lands at the PCB edges are located in a manner that enables them to be soldered direct to the relevant pins of the 25-way female (Ki) and male (K 2 ) sub-D con- Parts list Resistors (±5%): R 1 ...R 7 ind.-2K7 Semiconductors: D 1 ...D 7 incl.- bi-colour LED Miscellaneous: Ki- female 25-way sub-D connector. K2> male 25-way sub-D connector. PCB Type 890036 nectors (these are standard types with short, straight, pins, i.e., not special PCB-mount versions). It is recommended to fit the two bi-col- our LEDs for the RxD (received data) and TxD (transmitted data) reversed with respect to the other LEDs, so that a lit green LED always indicates a logic The final appearance of the RS-232 monitor depends much on individual taste. The completed board may either be cast in an ABS moulding, covered by cut-to-size metal plates, or built into an enclosure made from the hoods supplied with the D-connectors. These hoods are modified and then glued together to form a compact casing. Sound future for SMT Although there are still some who doubt the viability of Surface Mount Technology, there is ample evidence that the use of surface mount components is growing rapidly throughout the industrialized world. None the less, there remain a number of problems of which the most serious is probably the absence of agreed international standards of assembly and inspection. Another is the difficulty of visual inspection (automated inspection systems can not — yet — take over completely from the human inspector), which stretches human capabilities to their limit (think, for instance, of the thousands of solder joints on a single Eurocard). However, the first step to the solution of a problem is recognition of the problem and it is widely accepted that most pitfalls associated with surface mount technology have been recognized. In any case, the worldwide growth of SMT speaks for itself. If it were not a viable production method offering many advantages, it would have died a natural death by now. 7.50 elektor India July i98» SEMICONDUCTOR DIODES by T. Wigmore Although many readers know perfectly well what a diode is, it does no harm to repeat its definition here: it is any electronic device that has only two electrodes. There are two types of diode: thermionic and semiconduc- tor. The present article will discuss semiconductor types only. A semiconductor diode is basically a p-n junction, that is, a junction of n-type and p-type semiconductor material, currently usually silicon. An ideal junction of this nature, forgetting for the moment special types, such as zener diodes and varactors, behaves either as a short-circuit or as an infinite resistance, depending on the pola- rity of the applied voltage. Such a diode would possess differential resistance, rj, and d.c. resistance, R,\, only. Unfortuna- tely, ideal components do not exist and in a practical diode other parameters, such as bulk resistance, /?(,; junction capacitance. Cy diffusion capacitance, Cj; case capaci- tance, C c ; and terminal inductance, L, also affect its behaviour. These parameters are shown diagrammatically in Fig. 1. The deviation of a practical from an Fig. I. F.quivak'nt circuit diagram of a typi- cal small-signal or switching diode. ideal diode may be seen from the typical diode characteristic in Fig. 2. In the for- ward bias region, R j is fairly large until the threshold voltage is reached, after which it is small. In the cut-off region (note the different voltage scale), only a small (leakage) current flows in the diode until breakdown occurs, after which, except in the case of zener diodes, the diode is destroyed. Direct voltage When the voltage applied across the diode is direct or alternates very slowly, only R j, and /Jjj affect the behaviour of the diode: the other parameters in Fig. 1 may be ignored. The diode characteristic is then a function of the two resistances only. Since we can Fig. 2. Typical diode characteristic. Note the different scale of the -x and the +x axes. not deal with the derivation of the formu- las for these resistances in this article, we can only say that the threshold voltage in silicon diodes is 0.5-0.8 V and that in ger- manium diodes, 0.2-0.4 V. Once the thres- hold voltage is reached, the current would rise fast and linearly, were it not for the bulk resistance, which tends to impede the current, as can be seen in Fig. 3. In the reverse bias region, R^ is of little significance, since it is negligibly small compared with the conductance, Gj. The characteristic of a germanium diode is flatter than that of a silicon diode, both in the forward and in the reverse bias Fig. 3. Current vs applied voltage characte- ristic in the forward bias region with and without the effect of bulk resistance. Alternating voltage When an alternating voltage is applied across the diode, the various capacitances inherent in the diode (see Fig. 1) become the dominant parameters. Even at low-fre- quency voltages, these capacitances may make the diode unsuitable for certain applications. The relation between applied voltage, time and the consequent current through the diode is shown in Fig. 4. The junction capacitance is important for the behaviour of the diode in the rever- se bias direction, when a dense space char- ge exists at the p-n junction. At the instant the diode switches to reverse bias operati- on, the current through the junction capa- the current vs time curve and the voltage vs citance changes polarity (/ p to /p), and rapidly declines to a very low value (the leakage current, which is of the order of a few nanoamperes). The time it takes /r to fall from 90% to 10% of the value of /p is called the recovery time, t r When the voltage rises, Cj decreases exponentially, since the width of the space charge region increases. At zero crossings of the applied volta- ge, the diffusion capacitance, Cj, also affects the switching times, since the char Table 1 Type of diode Construction Properties Applications Alloyed junction Large cross-sectional area of barrier layer; large capacitances; high cur- rents; large tolerances Power diodes; zener diodes up to 10 V Diffused junction Large cross-sectional area of barrier layer possible; wide range of capaci- Power diodes; zener diodes above 1 0 V As diffused junction types tolerances; small dimen- sions and capacitances possible; good HF cha- General purpose; zener diodes; varactors; p-i-n diodes; Schottky diodes; HF diodes; switching diodes Planar epitaxial As planar types but with very tow forward resistance and very short recovery times Point-contact Very small capacitances; only small currents permissible; good HF characteristics General purpose (low reverse bias and low forward currents); HF (up to VHF region); switching diodes ge carriers in the semiconductor material have a certain inertia and act as short-term memories, particularly when a conducting diode is switched rapidly to reverse bias operation. It is, therefore, a requirement of rectifier and switching diodes that their diffusion and junction capacitances are minimal. At frequencies below about 100 MHz, the case capacitance and terminal induc- tance have but little effect, but as the fre- quency rises they become more and more influential and must, therefore, be inclu- ded in any computations. Types of construction The construction, properties and applicati- ons of five types of diode are shown in Table 1. Included in the table is the germanium point-contact diode in which, because of the very small contact area between anode (point) and the n-type germanium, the junction capacitance is very small (<1 pF) so that it is eminently suitable for high-fre- quency and fast-switching applications. The use of gold-gallium anodes allows switching times shorter than 1 nanosecond to be achieved. Also, its forward bias is smaller than that of silicon diodes. Against these advantages, it can not cope with cur- rents in excess of about 10 mA. Silicon point-contact diodes with simi- lar advantageous properties also exist, but because of their high vulnerability to over- loads they are not of great importance and are used only in very special applications. Germanium junction diodes have been superseded almost completely by silicon junction diodes and are nowadays used only where low forward bias is vital. Silicon junction diodes are produced principally by one of three methods. In the alloy process, the basic material is an n- type wafer of silicon doped with antimony into which an aluminium ball is inserted at high temperature. During the solidification process a sharply defined n-p region is for- med owing to the different fusion points of the materials and the diffusion of Si atoms in the aluminium. Because of the large area of the junction, this technique ensures that large forward currents are possible, although the device parameters are subject to wide tolerances. These tolerances are much smaller in the diffused junction process. In this, a wafer of n-type silicon with a very smooth surface is heated to 1300 °C in a diffusion oven after which its surface is changed to n + by a P2O5 dopant. Subsequently, the doping layer is removed from one side of the wafer after which this is doped with boron to make it p-type. The wafer is then provided at both sides with a terminal alloy after which it is sliced into small discs. The cross-sectional area, and thus the ensuing capacitance, may be given a fairly wide range of values. The diffused juncti- on process is particularly suitable for manufacturing power diodes and varac- Planar diodes are produced by a quite different technique. In this, a layer of sili- con dioxide, Si02, is thermally grown on the surface of a silicon sub- strate. Photolithography is used to etch holes in the oxide layer, which then acts as a mask for the diffusion of boron impurities to produce a p-type region. The cry- stal is then cut into small slices. This technique guarantees small dimen- sions, small capacitances and precise reproducibili- ty- Planar epitaxial diodes have an additional n + doped layer at the back which makes them extre- mely low-ohmic in for- ward bias operation. Schottky diodes are pla- nar epitaxial types wit- hout boron doping. Instead, they have a metal contact sintered direct on to the n-type substrate, which (because of the Schottky effect) acts as a p-type semiconductor. This has the advantage of greater hole mobility and, consequently, a smaller dif- fusion capacitance and shorter storage and switching times (about 100 picaseconds). Figure 5 compares the rectification of a 30 MHz signal in a Schottky diode and in a general-purpose diode. ges over a general-purpose diode for the rectifi- cation of a 30 MHz signal. Practical diodes After our short incursion into semiconduc- tor theory, we shall now look at some practical diodes. Small-signal diodes The most popular small-signal diode is the 1N4148. Although this has been around for about 15 years and costs next to not- hing, it has some very useful properties. With a parallel capacitance of not greater than 4 pF and a recovery time of 4-8 nanoseconds, it is eminently suitable for 7.52 Table 2 use in h.f. circuits. Its family includes the 1N4149; 1N4446-1N4449; 1N914A; 1N914B; IN916A and 1N916B, all with similar characteristics. A serious drawback of these diodes is their low forward cur- rent (max. 150 mA). Their reverse bias is of the order of 75 V and their dissipation around 440 mW. They are produced by the planar epitaxial technique. In applications where a low voltage drop across the diode is required, the Schottky types BAT81-83 (switching time <1 ns) or BAT85-86 (switching time <4 ns) are used nowadays, where in the past germanium diode Type AA119 would have been used. The Schottky types have a lower voltage drop (<400 mV), but their reverse bias of 40-60 V is lower than that of the AA119. Freewheeling and rectifier diodes For mains voltage rectification at currents below 1A, the most suitable diodes are found in the 1N400 1-4007 series. Their reverse voltage, depending on type, ranges from 50 to 1000 V. Apart from the fact that all diodes in the series are easily avai- lable, and at low prices, they can with- stand short peak currents of up to 30 A. For forward currents of up to 3 A, it is best to use one of the types in the 1N5400-5406 series, which withstand short peak currents of up to 200 A. Both series are manufactured by the planar technique. As an aside, a full-wave rectifier confi- guration using four discrete diodes is still cheaper than a proprietary bridge type. Fast freewheeling and rectifier diodes For operation at frequencies above 50 Hz, the diodes discussed above are too slow, and fast-recovery Types IN4933-4937 should be used. These are similar to mem- bers of the I N400 1-4005 series, but have recovery times of 100-150 ns. These times guarantee satisfactory operation up to about 250 kHz. They are typically used in switch-mode power supplies. Still faster are the BYV36A-36E series (reverse bias 200-1000 V; f r <100 ns); the BYV26/50-26/200 (1 A types) and the BYV27/50-27/200 (2 A types). The latter two series, all planar epitaxial types, offer recovery limes of not greater than 25 ns. High-voltage diodes High-voltage diodes are often encountered as rectifiers in cascode circuits. Their reverse bias is high — in the BY505: 2 kV and up to 24 kV in the BY741 . Diodes with low leakage current Diodes with very low leakage current are very hard to come by. Fortunately, they may often be replaced by good Schottky Type Typical parameters Applications 1N4148 Low forward current (200 mA; 400 mA max); last (4 ns); inexpensive Standard diode for small-signal and switching operation at low currents; free-wheeling diode for small relays BAT85 Low forward current; fast; inexpensive Schottky equivalent of 1 N4148; used in inductance and millivolt meters 1N400X Medium forward current (1 A); relatively slow; high peak currents up to 30 A Low-frequency rectifier; freewheeling diode; suitable for mains operation 1N493X Similar to 1 N400X but faster (1 50 ns); 1N4937 suitable for mains operation Fast rectifier; used in Elektor Electronics digital train decoder circuit 1N540X Medium forward current (3 A); otherwise as 1 N4001 Medium power rectifier BYV27 Very fast switching diode (25 ns); medium forward current (2 A); low reverse bias Freewheeling diode in stepper motor circuits; used in h.f. neon tube dimmers BYV26 Similar to BYV27 but at higher voltage and lower current (1 A) Used in h.f. neon tube dimmers BYV36 Similar to BYV26 but slower BYV79 Fast switching diode at high currents (14 A) Control circuits for radio control; used in 28 V converters BY VI 9 Schottky rectifier at high currents (10 A) Used in battery chargers types or. if really necessary, by a Type Schottky BYV19 may be used. The BF256B field-effect transistor of which BYV79 is particularly suitablefor use as a the drain and source terminals have been freewheeling diode. It can handle currents interconnected. of up to 14 A, has a reverse bias, depen- dent on version, of up to 20 V. Unfortu- Fast power diodes nately, it is not very fast (recovery time Fast power diodes are normally found in <50 ns) and has a voltage drop of 0.85 V power supplies whose primary circuits are at 10 A. clocked and in motor control circuits. Where these aspects are important, it is Suppressor diodes for operation at very better to use the Schottky version. This is high currents, such as the BZW86X not able to handle such large currents (up (12-85 V at 250-1000 A: dissipation 25 to 10 A), but its voltage drop of 0.6 V is kW) are not readily available and naturally significantly lower. Furthermore, its reco- tend to be very expensive. very time is only a fourth of that of the At lower powers, the BYV79 or the BYV79. MIDI SPLIT CONTROL A MIDI-compatible keyboard can be functionally split into a number of banks of keys with the aid of a straightforward computer program as shown below. The MIDI SPLIT facility dis- cussed is actually only a sub- routine from a purposely de- veloped MIDI control program written to run on a 6502-based microcomputer. As a source listing is given as part of this article, the MIDI SPLIT routine can be studied in detail by pro- grammers whose micro allows them to write object code direct into the memory, or through an assembler. The computer should be equipped with a Type 6850 ACIA (Asyn- | chronous Communications Interface Adaptor) pro- grammed to send and receive MIDI data at the standard baud- rate of 31.25 K. For the ACIA to operate at this data transfer rate, its clock input must be 500 kHz. A MIDI interface must, of course, be fitted at the serial I/O port of the computer. The proposed machine language program resides in less than two pages of RAM, and may need a patch here and there to make it run on a par- | ticular system. An easily writ- ten BASIC program could be added to read the desired SPLIT POINTS. Provided you are sufficiently well acquainted with the inter- nal memory organisation of the micro in question— it helps when you have built it yourself —this MIDI SPLIT subroutine can offer features not com- monly found on even the most I expensive of programmable | MIDI keyboards. To begin with, the number of split points that can be used to define the size of the banks of keys is not limited to a mere three or four; this program actually supports the use of up to fifteen user- definable split points. Each of the banks can be arranged to control several MIDI channels, the minimum number being nought (this is definitely not in- sensible), the maximum num- ber four. The control of more than four MIDI channels by a single bank of keys is prob- lematic because this lays rather a heavy claim on the accepted data transfer rate of 32 Kbaud. In essence, these 350 or so bytes turn your computer into a MIDI SPLIT PROCESSOR in- serted in the data path from the MIDI keyboard to the relevant input of the synthesizer or any other MIDI-compatible musical instrument. This means that your keyboard henceforth func- tions as a MASTER KEYBOARD with the previously mentioned exceptional features. Import- antly, the proposed program is fully transparent to the VEL- OCITY parameter. Interrupts for speed It will be understood that the | | proposed program must be so fast as to ensure that the data I stream from the keyboard to the instrument is not in any way slowed down. It is, therefore, hard to get round a factual im- plementation in an interrupt- based structure. Unfortunately, , the proper dealing with inter- I rupts is a major headache for many programmers, whose re- sulting low spirits are often caused by the INT line in the system being low at the same time. In order to avoid difficulty arising from it being incor- porated in the computer's inter- rupt household, the present \ program has been kept fairly 5 simple and purposely does not make use of the 6502's zero- page. As shown in Fig. 2, the ex- ecution of the MIDI processing routines can be interrupted by an IRQ pulse from the ACIA, whenever its receiving register is filled with MIDI data from the keyboard. Before examining the program in greater detail, it is important to understand that it first filters the incoming MIDI datastream, then compares each MIDI word with entries from a table that holds information about the key j numbers representing user- defined split points, and about the bank-to-channel assign- ment, and lastly outputs the data to the appropriate MIDI channels. It is readily seen that the use of an intelligent data routing device sitting between a MIDI keyboard and an electronic musical instrument offers to the user a whole new scope of in- teresting and quite sophisti- cated MIDI data processing methods such as transposition, octave-shifted accompaniment, fifths and thirds, selective sup- 1 1 1 1 J SFF SFF Iff 3rd channel SFF SFF SFF SFF SPLIT 2nd channel 4th channel '““”"1 ' 1 ' 1 1 1 ' 1 ■ 1 SS 1 1 1 1 ^ channel (CHANEZ) j 2 | 2 | SFF | 2 | SFF | SS 1 1 1 1 <“"** 1 > 1 i 1 SFF | m I »F I SS 1 1 1 1 (CWM | 4 | SFF | SFF | $FF | SFF | SS 1 1 1 1 Fig. 1. An example of how split points are brought into effect to have banks of keys control specific pressibn of MIDI data (velocity, Program description key to enable halting the The fust entry in this table must after touch) which would other- With reference to the flowchart program at any time without the be the rightmost split point, wise cause "partially compat- in Fig. 2, and the source listing, need for a general system as shown in Fig. 1. CHANEi is ible” instruments to produce Table 1, it is seen that the ACIA RESET. also a look-up table, but its undesirable sound effects, occupies two addresses: one SPLITS is a variable that holds 16x4 bytes are reserved for the the adding of a software-im- for its command register (at the number of desired split channel numbers that go with plemented sustain or soft pedai, $E120), and one for its data I/O points, while SPLIT is the label each bank of keys defined with and so forth. For the moment, register (at $E121). The CON- for the 16-byte table in which a set of split points. Any nega- however, we will concentrate TROL C function of the ASCII the user has entered the key tive value— i.e., one greater than on the MIDI SPLIT function. keyboard is used as the BREAK numbers that mark a split point. 7Fh— marks the end of the series of channel entries. The contents of FLAG serve the double function of information status indicator and key code already received flag ($80 = key off; $90 = key on; 0 = key number already received). IRQPNT is the pointer for the IRQ FIFO (first-in first-out) stack. KEYNMB holds the number of the key whose command is cur- rently processed, and VELOCT is a byte that holds the cor- responding key depression speed (bit-manipulation on this byte may be used to bring a software-supported soft pedal into action). STAT indicates whether a block of MIDI data currently processed originates from an activated or a released key. CHNCNT is a variable set up for the counting of the MIDI channel numbers that go with each bank of keys. The Y register in the 6502 func- tions as a read vector for the IRQ stack, and must not be con- fused with IRQPNT which con- trols the write actions. Begin with BASIC The simplest method of pro- viding for the split point and channel assignment codes in the machine language program is the running of a BASIC program that prompts the user to input his set of parameters before the actual MIDI SPLIT routine is called into action. The desired values are POKEd into the appropriate address reserved for SPLITS (number of split points), the address range SPLIT. . .SPLIT +15 (key numbers that mark a split point), and address range CHANEI... CHANEI +63 (cor- responding channel numbers). With some skill in machine language programming, a sub- routine could be written to ef- fect the loading of a new set of parameters at the touch of a specific key on the MIDI key- board, rather than one on the Fig. 2. This flowchart of a MIDI control program illustrates how the machine language routine on computer, the previous page is used as the basis for further experiments in the writing of MIDI software. DM direct driving of auxiliary syn- thesizers. This means that auxiliary synthesizer no. 1 and 2 receive an identical input signal, and hence are correctly synchronized under all cir- cumstances. The above discussion should not lead to the conclusion that the quality of a MIDI instrument can be judged from its number of input and output sockets. As set out above, a long chain of series connected MIDI instru- ments readily leads to trouble- some asynchronicity, owing to the incurred phase delays and pulse distortion. The MIDI | redistribution circuit proposed here provides the means for controlling a large number of instruments from the main syn- thesizer, without running into difficulty as regards distortion of the serial MIDI signal. The redistribution unit is a relatively simple circuit, which can be built by anyone capable of cor- rectly soldering 5 wires to a 5 way DIN plug. 16 MIDI outputs The use of the MIDI redistribu- tion unit is illustrated in Fig. 2. Note that the instrument con- figuration shown is but an example; other uses of the redistribution unit are feasible, as will be seen below. The circuit diagram of the MIDI redistribution unit appears in j Fig. 3. The four inputs are stan- | dard MIDI types, i.e., based on j the use of an optocoupler. The Type TIL111 is an inexpensive and commonly available opto- coupler, but its electrical per- formance is not spectacular— the MIDI signal is typically delayed by about 9 ^s, and the duty factor is altered con- siderably. None the less, the device gives satisfactory results in this circuit. For those con- structors striving towards near perfection, the design of the circuit board allows the fitting of the fast optocoupler Type 6N13S. After reshaping and inversion of the incoming pulses in gates Ni, N?, N13 and Nio, the signal can be distributed in various ways over the 16 available DIN output sockets, each of which has a standard current loop in- terface. The four remaining inverters Ne, N12 , Nu and N24 are connected to function as LED drivers for the four inputs of the circuit. NEW PRODUCTS INSTRUMENT CABINETS System Engineering brings new concept to the electronic industries in the field of instrument cabinets. The instrument cabinets are suitable for standard DIN panel cutouts. These are suitable for in- strumentation industries, R&D centres, educational institute, test and measuring instruments, etc. SE-44, SE-63, SE-84, SE-42, SE-33, SE-66 and SE-88 are the various instrument cabinets to suit panel cutout of 92 x 92 mm, 138 x 67 mm, 186 x 92 mm, 92 x 45 mm, 67 x 67 mm, 138 x 138 mm, and 186 mm x 186 mm, respec- tively. Each model is available in 80, 120, 160, 200, 250, and 300mm depth to cover entire applications. For more details write to: SYSTEM EN- GINEERING, • 38-39, Hadapsar hull. Estate, • Pune, • Maharashtra 411 013 • Phone: 670962, 671951. SMPSfprB/W Television ABR Electronics have developed six models of switch mode power supplies (SMPS) for B/W 51 cm television sets. The Series-A SMPS models have mains isolation and short circuit protection. These give the desired DC output vol- tages for AC input variation between 90 V and 270 VAC. Marketing Division • ABR Electronics (P) Ltd. • Srinath Complex • First Floor • S.D. Road • Secunderabad-500 003. Potentiometric Strip Chart Recorder PROTEK LM 120 R potentiometric strip chart recorder uses DC linear ser- vomotor principle, which gives long- term reliability and comparatively better performance. The linear servomotor provides the pen-drive across 120 mm calibrated scale. The disposable pen gives a single, continuous, smudge-free trace, avoiding spreading and spilling problems. Different coloured pens are available. Plug-in chart cassettes signal of 0-10 mA or 4-20 mA as option is also available. The recorder can record al- most any process variable like tempera- ture, pressure, weight, humidity, con- ductivity, pH, %, or Co 2 that can be translated into an electrical signal. 3, Parvati • Janaki Apartments • Chin- tamaninagar • Sahakarnagar No. 2 • PUNE-411 009. Ioniser & Air Purifier THE Ioniser and air purifier can be used to clean the air of pollution and balanc- ing the positive and negative ions per- centage in the air. Negative ions work as “AIR VITAMINS’’ and can give relief to ASTHMA PATIENTS as well as aller- gic patients. The Ioniser can be used in bed rooms. Kitchens, drawing rooms. offices, hospitals, operation theaters, factories etc. The unit is enough for a room size of 1500 cubic feet. It works on 230 VAC and power consumption ap- proximately 1.5 W per hour. M/s. Prayag Electronics • A-5, Success chambers • 1232, Apte Road • Deccan Gymkhana • PUNE-411 004. Time Switch THE MIL 2008 Q series is fitted with a quartz electronic drive control and a step motor. The quartz frequency is 4. 19 mill- ion Hertz and the quartz stabilisation en- sures the exact running of the driving mechanism. These time switches are de- signed for the accurate and effortless control of oil heating installations, elec- tric heaters, airconditioning plant, water processing plant, street lights, traffic sig- nals, etc. MIL 2008 Q is available with contact rating of 16 A, 250 VAC and with daily programme and weekly prog- ramme dial. It operates on mains supply and continues to run for 150 hours after power failure on a battery back-up. M/s. Sai Electronics • (In association with cupwud Arts) • Thakore Estate • Kurla Kirol Road • Vidyavihar (West) • Bombay-400 086. • Ph: 5136601/ 5113094/5113095. NEW PRODUCTS resistance, eliminates errors due 7 mm LED display and operates on 230 V± 10%. Temperature Programmer SCR Elektroniks have developed a temperature programmer, the Model Step Prog-8, for use in any industrial or research process requiring accurate temperature control at different temper- ature setting levels for pre-determined times. Eight levels could be set at the be- ginning of the process. Similarly corres- ponding time periods can be set for each temperature level. The accurate temperature is indicated on digital temperature indicator. The step- in progress is indicated on LED marked Program number. The temperature set- ting of that step can be read on digital display by pressing “Press to read” Pushbutton. The facility to actuate an alarm for lower or higher temperature is optionally available and a delay timer is provided to silence (mute) the audio alarm for a period settable with knob (potmeter). Battery back-up is provided to maintain the programme and the current step in fact in the event of power failure. M/s. SCR Elektroniks Opp. Fatima High School • Kirol Road • Vidyavihar • Bombay-400 086. Tel: 5134605/5128057. Digital LCR Meter INDUCTANCE, Capacitance and Re- sistance measurements can be done by the Vasavi Digital LCR Meter by eliminating bridge balancing. Display of Simultaneous Tan Delta (dissipation fac- tor) facilitates checking the quality of the component. The ranges covered are 0.0001 ohm to 20 Mohm, 0.1 pF to 20,000 mF, and 0.1 micro Henry to 2000 Henry. For lOw capacitance measure- ment Guard Terminal is provided to eliminate measurement error due to stray effects. Four Terminal measure- ments for large capacitance and very low M/s. Vasavi Electronics • (Marketing Di- vision) • M7 & 8, Chenoy Trade Centre • Park-Lane • Secunderabad-500 003. • Phone: 70995. Agrawal Sales Enterprises • 34, Ganesh Bazar • Jhansi-284002 (U.P) Digital Time Interval Meter PLA digital time interval meter has a measurement range of from 1 m sec to 9999.9 sec. Its 4 or 5 digits LED display gives high accuracy for smallest time in- terval measurement of 0.1 m sec. It has a provision of measuring time interval in 16 different modes. The Time interval meter is used for measuringh Switching time of relays, trip time and On time of circuit breaker, hav- ing time of fuse element, travel time of switch and contractor, etc. It can be used for measuring the time interval by step positive (2 V to 5 V DC) Voltage applica- tion/removal mode. Soldering Station The Model DAA-IO soldering station features a temperature sensing element positioned at the tip of the soldering iron which continuously senses the tempera- ture at the iron tip and sends correspond- ing signal to operate the relay supplying to the heater element. The heater of op- erates with regulated 24 VDC stepped down from 230 VAC 50 Hz through an isolation transformer. The irorf tip is per- fectly grounded for the safety of the com- ponents which are to be removed. The unit is suitable for soldering components like ICS. CMOS etc. M/s. Pla Electro Appliances Pvt. Ltd. • Thakor Estate • Kurla Kirol Road • Vid- yavihar (West) • Bombav-400 086. • Ph: 5132667/5132668/5133048. M/s. Electro- Vac • 10/430-431 • Tasveer Apartments • Soni Falia • Panini-bhint •Surat-395 003. Opto Digital Capacitance Meter OPTO Digital capacitance meter incor- porates high quality integrated circuits so that reading is not affected by the capacitor leakage current. The instru- ment has various ranges to measure dif- ferent values of capacitors with 3 Vis digit 7.64 ele NEW PRODUCTS Plug-In Modules & Instrument Cases. EUROPACK have developed what is said to be a new concept in 19” plug-in units (sub-racks) and bench top model instrument cases. These cabinets are of sturdy frame construction, using aluminium extruded sections. The Ms cover plated are inserted through the slots in the depth extrusion. The handles are designed to take heavy loads. Four feet give the bench case facility, which the two front feet have tilt facility. Spe- cial dies give ventilation slots. No screws are visible on the cabinet except 4 Nos. On the rear panel, covered by moulded plastic washers which also act as legs. These cabinet also can be made as per IP53 specifications. The standard heights are from 2 U to 9 Us (1U-44.45 mm). The standard depths are from 150 mm to 500 mm in multiplies of 50 mm. The standard widths of the bench cases are 84 TE , 63 TE , 56 TE , 42 TE, 28 TE, and 21 TE (1 TE -5.08 mm). Apart from the standard sizes, cabinets can also be made as per requirements from standard components. M/s. Europack • C/10, Laghu Udyog Kendra • 1 B, Patel Road • Goregaon (E) • Bombay-400 063. Inductive/Capacitive Proximity Sensors HANS Turck GMBH & Co. KG (feder- ation Republic of Germany) manufac- ture inductive proximity switches with sensing distance of 60 mm, and capaci- tive proximity switches of 10-40 mm. Special proximity switches are available for explosion, welding pressure and high temperature-proof applications. Pther products are motion control gear, in- cluding rotational speed monitor, rota- tional speed meter, speed sensor and di- rection discriminator. These device also monitor other repetitive movements. M/s. Aran Electronic Pvt. Ltd. • 2 E, Court Chambers, • 35, New Marine Lines • Bombay-400 020. • Tel: 252160/ 259207. Digital Line Frequency Meter ANU Vidyut Digital Line Frequency Meter Type 321 is for measuring line fre- quency in power plants, sub-stations, distribution centres, etc. High accuracy and long term stability is made possible by incorporating a crystal controlled clock generator. The measur- ing frequency frequency remains 50 Hz to 99.99 Hz with accuracy of ± 1 digit, operating voltage in range of 180 V - 280 V AC single phase. A clear 25 mm LED display clearly indi- cates the frequency. The instrument is lightweight and suitable for panel DIN 144 mounting i.e. panel cut-out to be 135 x 135 mm; also available in DIN 96. Anu Vidyut • C-l, Industrial Estate • Roorkee-247 667. TOTALISER JELTRON offer the Model 810A mic- roprocessor based digital indicator-cum- totaliser suitable for a variety of indust- rial applications. The front panel con- sists of four-digit LED display alongwith a user friendly membrane keyboard. The totaliser based on 6502 microprocessor, can be used areas like bas and liquid flow totalising, KW hour, totalising and so on. Engineering units i.e. litres/hour, litres/minute, decimal point positioning for ranging, and totalising update time are all programmable throught front panel keyboard. The totaliser accepts an analog input sig- nal of either 0 to 5 VDC, 0 to 10 VDC 4- 20 mA current, or millivolts input. At any given time the totalised output can be seen by using the front panel keyboard. Similarly it can also be reset using the front panel keypad. M/s. Jeltron Instruments (India) Pvt, Ltd. • 6-3-190/2, Road No. 1 • Banjara Hills • Hyderabad-500 034. 7.66 NEW PRODUCTS Illuminated Magnifying Glass MARVEL Products offer the M-Plast il- luminated magnifying glass for en- gineers, scientists, type-setters, finger print experts, proof readers, bankers, ar- tists, hobbyists etc. The 23 cm long plas- tic moulded body has 9 cm round mag- nifying glass. Hi-beam light can be switch operated. It works on four 1.5 V penlite batteries. M/s. Marvel Products • 208, Allied indl. Estate • Mahim • Bombay-400 016. • Phone: 468346/466846. Emergency Lamp EL 636 is a portable emergency lamp fit- ted with 8” flourescent single tube, and the automation is fully electronic. The storage cell is maintenance free, and the body is of fibreglass reinforced plastic, the lamp can continuously work for three hours. Built-in invertor is heavy duty, long life. Dimensions are H-9”, L-2.5”, D-4”, and weight 1.4 kg. M/s. Transworld Electronics • (Market- ing Division) • 26/571, Oottukuzhy • Trivandrum-695 001. Electronically Temperature Controlled Soldering Bath An Electronically controlled tinning bath suitable for uniform and perfect tin- ning of delicate electronic components has been developed. It works with input voltage of 230 V ± 10% with output vol- tage of 0.75 KVA. The Temperature can be set between 170° and 350°C. The capacity of the bath will be around 500° M/s. M.R.K. & Brothers Engineers • 310 A, Commence House • N.M. Road • Fort • Bombay-400 023. Electronic Conveyor Belt Weigher ENCARDIO-RITE's Model ECBWS- 101 Weightoveyor is a precision elec- tronic conveyor belt scale designed to continuously weight any bulk material that can be conveyed, indoor or outdoor in dusty or wet environment. It is con- structed from heavy duty structural steel to permit complete torsional stability. The design brings the total sensed weight to a single point so that it can be moni- tored by a precision strain gage type of load sensor, Encardio-Ritc's Model EAU-310 load cell. The Weightoveyor offers an accuracy of ± 0.25% fsd for 4 idler systems, and ± 0.5% fsd for 2 idler systems. M/s. Encardio-Rite Electronics (P) Ltd. • A-5, Industrial Estate • Talkatora Road • Lucknow-226 Oil (India) • Tel: 50382, 52130. Moisture Meter OPTO 1100 M Series of portable mois- ture meters is for quick and accurate de- termination of percentage moisture con- tents of organic and inorganic materials as well as hygroscopic materials, such as timber, soil, cotton, grain etc. The in- strument operates on pencil battery cell or DC power supply. The results are independent of any varia- tion in ambient environmental condi- Agrawal Sales Enterprises • 34, Ganesh Bazar • Jhansi-284 002. Capacitor Holder NOVCFLEX have developed a one- piece holding device for fixing and hold- ing aluminium can type electrolytic capacitors. Application is simple:- just position it on the capacitor and press the jaws together. For maximum tightness a plier can be used to give that little extra firmness and vibration resistance. By ap- plying lateral pressure, the snapper capacitor holder can be easily released and refused again and again. The holder is made from high engineering thermop- lastic polyamide displaying high strength, toughness, flexibility, excellent abrasion resistance and good electrical insulation. M/s. Novoflex Cable Care Systems • Post Box No. 9159 • Calcutta-700 016 • Tel: 29-4382, 29-5939, 29-3991. NEW PRODUCTS Precision Digital Multimeter PREM A (Prazision Electronic und Mess Anlagen GmbH) of Fed. Republic of Germany offer seven high accuracy 6 Vi- digit resolution DMMs in a range. The top-of-the-line-DMM 6031 A has a ohms stability of 2 ppm for 24 hours, and ac- curacies of 0.07% for AC volts, 0.005% for Dc and 1% for AC currents. Temper- ature tolerance is 0.05°C IEEE 488 bus interface is a standard feature. DMM 6031 A has a 10 gigaohm input resis- tance. A series rejection of more than 100 dB is attained because of the inhe- rent advantages of PREMA's patented multiple ramp integration synchronised by PLL to the mains frequency, and ad- vanced shielding techniques. The DMM can be fitted with an inbuilt 20 channel 4 pole scanner (thermal offset 1 /xV) for use in multi-point measuring systems. It has a wide scope of data processing oper- ations on the measured values using its set of 20 mathematical programs. Func- tions include 8th order polynominal linearisation, non-linear, trigonometric, and statistical functions, etc. Up to four programs can be cascaded in any desired sequence to give a new compound prog- Electronics Engineering Services *231 Keytuo Industrial Estate • Kondivita Road • Andheri (East) • Bombay-400 059 • Digital pH/mV Meter Puneet 3'A digit PH meter model No. PH-1 ID is a LED type portable instru- ment for laboratory R&D, and educa- tional institutions. It has extremely sta- ble DC amplifier with high input impe- dance. It provides manual and automatic temperatures compensation in the range of CPC to 130°C. It has asymmetry ans slope correction controls for periodic caliibratiop. It also provides recorder output and titration facilities. It mea- sures pH from 0 to 14 and mV from o to ± 1999 mV with automatic polarity and over-range indication. The unit is housed in an elegant moulded cabinet and weights less than 1.5 kg. M/s. Puneet Industries • H-230, Ansa In- dustrial Estate • Saki-Vihar Road • Bombay-400 072 • Stroke Counter CE Industries offer the 5 digit stroke counter model No. CSO30 with large display. A knob reset facility brings all the figures to zero. No lubrication is re- quired as all the moving parts are made of self-lubricating material. The counter is used for printing preses, duplicating machines, circuit breakers, power pres- ses, injection moulding machines, etc. M/s. Sai Electronics • (A Divn. of Starch & Allied Industries) • Thakor Estate • Kurla Kiron Road • Vidyavihar (West) • 400 086 • Ph: 5136601/5113094/5113095 In-Circuit Tester Kandenstsu Ltd. of Japan, offer the Fussa, Cabol 3301, a parts mounted board tester that helps accomplish three critical functions viz. precision in mea- surement, test speed and analog isola- tion, in a well balanced manner. The 3301 offers 320 test points, expandable to 1024, in steps of 32 points. Measuring speed for short test is 3 seconds/320 pin. Maximum measuring steps are 2048 (each measuring step tests a component) at the speed of 15 ms per step. COBOL 3301 features automatic guarding facility which not only simplifies complex mea- surements but also uprates the measure- ment precision. The maximum number of guarding points for each test step is 15 with guarding points for each test step is 15 with guarding current as high as 100 mA. The measuring range for Cobol 3301 covers, resistance 0.1 ohm to 100 Mohm, capacitance 1 pF to 100,000 mF, inductance 1 uH to 100 H, diode and transistor 0.1 V to 2.0 V and Zener Diode up to 40 V. Applicable PCB mea- surement is 450 mm x 350 mm (maximum). The Fussa Cobol3301 range consists of: Gorilla for press type fixture, Elephant for vacuum type fixture, and Dragon automatic feed in-circuit board tester. HCL Limited • Instrument Division *G- 5 & 6 Vaikunth *82-83 Nehru Place • New Delhi-110 019* 7.70 CORRECTIONS Pitch control for CD players January 1989. On the component overlay of printed-cir- cuit board 880165 (Fig. 7), the capacitor next to C21 should be marked C20, not C19. The value remains the same at 100 nF, but a ceramic capacitor should be used as ad- vised in the Parts List. Colour test-pattern generator January 1989. Diodes Du, D17 and Du are shown with the wrong polarity on the component overlay shown in Fig. 5. Autonomous I/O controller (part 1) January 1989. Table 1 should be inverted: no diodes fitted gives instrument address 150-151, and both diodes fitted address 144-145. The digital model train (part f) April 1989. In some cases the operation of the locomo- tive decoder is affected by points control commands. This problem can be solved by increasing the value of Ri from 12 kfl to 39 kfl. The circuit diagram (Fig. 16) should be amended accordingly. LFA-150: a fast power amplifier (final part) January 1989. On the component overlay of the protec- tion board shown in Fig. 10, the plus sign at the negative pole of electrolytic capaci- tor C«s should be removed: the printed capacitor symbol indicates the correct po- -x Allowed lo post without prepayment. LIC No. 91 MH BY WEST-228 LIC No. 91 Get the ECIL advan tage! Tap the dedicated design group to develop the working hybrid you need Besides the size and weigh! reduction for enhanced portability. Hybrid Microcircuits tree you from chores Involving layout design, procurement ot components, inspection, inventory planning, population and final testing. A series of steps that invariably go inlo the making of o microcircuit. ^ EClLs dedicated design group jMp aided by CAD facility can help JVv/ you develop the working •'■'d'', hybrid you need Be It in telecommunications or computers or what have you mRc m The manufacturing W facilities are about * the best anywhere: Z f High speed screen printing machines, computerised ~?X ~' automatic loser trimming system, high-speed IP automatic pick and place system, automatic functional testing equipment, etc. All of which add up to give you the advantage typically EClLs Technical or commercial ECIL Hybrid Microcircuits §< rz/d rrnirrni vrrwrm urr ifvTur fMf va » ~ ^Vil' Live Electronics Corporation of India Limited »