STAI BS From the Pliblishers of ETI &HEBritain's Biggest Magazine For The Sinclair User Over 100 pages of information and programs for the ZX81 and ZX80 user including a 1K Chess routine! ZX Computing is a quarterly magazine with over 100 pages of programs, ar- ticles, hints and tips for the 2X81 and ZX80. We've put the magazine together with just one aim in mind — to make sure you get the most use a! out of your Sinclair computer In the'first issue, as well 2 dumped directly onto the p mysteries of machine code, s routine which takes up just IK, an ex- planation of PEEK and POKE, an entire section on business uses of the ZX81 (complete with a program to handle the ac p to 100 customers) and a construction article which will show you — even if you've never used a soldering iron before — how to cheaply double the memory on your ZX81 If you're serious about exploring the full potential of your Sinclair com puter, and you want an easy-to-understand series of articles to improve and develop your programming skills, then ZX Computing is for you. Issue one is on sale now d enjoyment a host of 1K and 16K programs (the majority > articles to help dispel theEditorial and Advertisement Office: 145 Charing Cross Rd, ‘London WC2H OEE. \veriement Manager: a in Griffiths Menaging Director: Ti. cennel speciale Publications Ld, 18 Special Publcttons Ltd, oe ‘148 Charing Cross Road London WeaH OEE DISTRIBUTED BY: Argus Press Sales & Distribution fe 12-18 Paul Street London is (British eles) tay Sea a tay pauy He ayy He i FHT il ue ae Ba ‘telephone. Vol. 3 No. 1 Summer 1982 Designer's Notebook i : i CHAPTER TITLE PAGE 4 1 Impedance and Phase 4 2 Anatogue Delay Lines 9 3 Radio Principles 13 4 Iwrline Switching 22 5 Time Constants 26 6 MPU Support Circuitry 26 i 7 Analogue Switchin 30 i 8& Batteries Explained 37 yy 9 Revisited 43 10 VL Techniques 56 11 CMOS: 3550 Applications 49 12 Television Principtes 56 13 Op-amps and Integrators 61 16 Eddy Cmrents 65 15 Low-level Audio Design 67 16 Atomic Structure 73 f 17 Bandpass and Beyond 76 i 18 Uncertain Principles 82 | 19 Magnetic Amps 84 20 The 1537 UCA 87 21 Digital Design PrinciplesChapter 1 IMPEDANCE AND PHASE Life would be a lot easier if all components behaved like resistors. Inductors and capacitors make life difficult by separating voltage and current, so how do you find the voltage or current at any point in a circuit? Phase diagrams to the rescue. IN ELECTRONICS one often needs to know what the voltage of current at some part of a circuit will be, without actually building it to find out. When dealing with DC, this is usually pretty straightforward, using Ohm's law and a few rules of thumb. But AC signals in a circuit are a different matter, often reacting in totally different ways, predictable only by using impedance theory and phase diagrams. It is this type of theory and the calculations used to find voltages, etc., in circuits, that concern this article, AC Signals First let's remind ourselves what an AC signal actually is. Plotting voltage against time for a typical signal would give us a graph like that in Fig. 1. This particular variety of round wave is known as a sinewave, and in order to fully describe it we must outline two quantities: its rms value and its frequency. The former is a measure of the amplitude, or height of the wave, and for reasons that need not be gone into here, is, in the case of a sinewave, 0.707 times the ‘maximum value of the wave. For instance, if, as in Fig. 1, the wave has a maximum value of 6 volts, the rms value of the signal is 0.707X6=4,24 volts. The other measure of the wave is the frequency. Take the interval between, say, A and B on Fig. 1. This interval, from one point to the next point where the voltage is acting in exactly the same way (in this case, from a point where itis zero and decreasing to the next point where itis both zero and decreasing) is called the period of the wave and is measured in seconds. During one period, the wave is said to have gone through one full cycle. ‘The frequency of the wave, we can now say, is the number of cycles per second. Fig. 1. A voltageritime graph fora typleal AC signal. Im Impedance can be described as the opposition to electrical current given by a circuit. Of course, we know about ordi- nary resistance, but there are other varieties. For instance, a capacitor may have a very high opposition to DC current, ‘but a very low opposition to AC signals of a suitably high frequency. This obviously isn’t ordinary resistance, because if it was, it would remain the same for AC and DC. In fact, the amount of opposition given to a signal by a capacitor is ‘measured by the ratio of voltage across it to current through it (Vil. This ratio is called the ‘capacitative reactance’ of the component, and is given the symbol X... Like resistance, reactance is measured in ohms. Capacitative reactance may be calculated from the value of a capacitor by using the formula X.=1/2nfC, where x is the Greek letter Pi, and 2. Current and voltage plots fr a capacitor, showing a phase between the to epresents the number 3.14, Tis the frequency of the signal being applied, and C is the value of the capacitor in Farads. Note that, as stated earlier, the opposition (re- actance) of the capacitor becomes very small at high fre- quencies, but to DC (where the frequency is effectively zero) or to very low frequency signals, it becomes effecti- vely infinite. Inductors, too, have a variable reactance; in this case, the inductive reactance, X;, which may be obtained from the value, L, in Henries of the inductor, from the formula X=3atL.. Note that this reactance also varies with fre- quency, but here, it becomes greater at high frequencies, approaching zero only when f is very low, ot when DC is encountered. Again, Xp isthe ratio Vil in the inductor, and thus, given either the voltage or the current, itis possible to calculate the other in either a capacitor or an inductor, if we know the frequency at which the circuit is operatingTo conclude this section, we now give a rather more adequate definition of impedance than that which we began with. Impedance is the combined opposition to AC signals in a circuit given by the resistance and reactance of the circuit. If we represent it by Z, the resistance by R, and the reactance by X, then Z=VR7+X", We find that, in a combination circuit of several components, Z=V/1. Phase Differences In addition to information about voltages and currents in circuits, phase diagrams also give us information about phase differences in these circuits. What in the world is a phase difference? To answer that, we must return to the ‘capacitor and inductor. Suppose that we are applying an AC voltage across a specimen of the former type of compo- nent. If we now look at the current flowing through it, we find that itis ‘leading’ the voltage by a quarter cycle. That is, although it goes up and down in the same way that the voltage does, the two quantities are not in time with each other. If the voltage has, say, gone up (as from point A to point B in Fig. 2), then the current did so XP, or a quarter ‘of a cycle earlier. (The figure 90 is used because a full cycle is taken as being divided into 360 degrees, as a circle is, and fone quarter of a cycle is therefore, represented by ¥%46x360=90. The reason for dividing a cycle into 360 de- ‘grees will become apparent later). If we superimpose a graph of current against time on top of one of voltage against time, we get something like Fig. 2. In the inductor, a similar effect occurs, but here it is voltage which leads current by 90°, rather than vice versa. 3. A series o an inductor. Do you use voltage oF i. 3, Aes rel witha estar nda induct, Do you we vag ‘The ‘phase difference’, as it is called, is given both in the ‘case of the capacitor and the inductor, the symbol @ — the Greek letter Phi — and may also be measured in terms of radians, another unit of angle, rather than degrees. To help remember that voltage leads current in the indi- cator, whereas current leads voltage in the capacitor, the mnemonic CIVIL is used. In a capacitor, (C) current (1) leads voltage (V), but voltage leads current, (I) in an induc- tor (L). Taken in order, the one-letter symbols for the components, voltages and currents spell CIVIL. (All right = Ididn’t think of it.) Phase So far we have seen how voltage and current are related in terms of magnitude (size) and phase, in individual compo- nents. What happens, though, if we put two different com- ponents ~ a resistor and inductor — for example, in series r parallel? This is where the phase diagrams step in, folks. Phase It Let us suppose that these two components, each of known value, are connected in series, and that we know the cur- rent which is flowing through the combination, and this current’s frequency. We wish to find the size and phase of, the total voltage across the two components, and we might be misled into thinking that it would be just the sum of the two individual voltages across the individual components, but in fact, this will not be so. The current and voltage will be exactly “in phase’ in the resistor, but in the inductor, the voltage will be 90° out of phase; you can’t just add voltages unless they are in phase with each other. Of course, we could find the magnitude of the total voltage by finding the total impedance of the circuit and multiplying this by the current, but we still wouldn’t know the phase of this voltage with respect to the current, so a phase diagram is really our only option: 4 The vliges crs the resistor and Indvctor can be ated to ind the {ota ons serie fe compete Which Reference For our diagram, we shall want some quantity, either voltage or current, which will be the same for both compo- nents. Well, as we have just seen, the voltages across the individual components are definitely different, so that only leaves current. In fact, current serves as our ‘reference ‘quantity’ in any series circuit, and voltage is used in parallel circuits. To represent the current, draw an arrow, pointing to the right. Now we must draw in arrows to represent the voltages across individual components. The lengths of these arrows will be made, using a suitable scale, to represent the sms values ofthe voltages. The phase of each voltage with respect to the current will be indicated by the angle, goi tnitelockwise, which the volage’s arrow maker wi tnt of the current, when both have their tails at the same place. Thus, the voltage across the resistor, which can be calcu- lated by multiplying the current by the resistance, will be represented by an arrow actually on top of that showing the current, because the voltage and current here are in phase, so that the angle, 9, is zero. The voltage across the inductor ‘ean be calculated by finding the reactance of the compo- rent, and multiplying this by the current. This arrow will be placed at an angle of 90° to that representing the current Ge, it will point straight up), because the voltage in an inductor leads the current by 90°. Were the component a capacitor, @ would be —90°, because the voltage here lags by a quarter cycle, which is equivalent to saying that it leads by —90°. The arrow would, then, point down, rather than up, as it does now.Sa, Aresistor and an inductor nth case voltage is sed as ‘hectatac quant, 0) feepaae top foraparaie eda Tf we imagine our two voltage arrows to be two sides of a parallelogram (in this case, a rectangle, because we know that one of the angles is 90°), and draw in the other two sides parallel to the ones we have, as in Fig. 4, we find that, the diagonal of the rectangle, drawn in as an arrow starting at the same place as do all the others, has a length that, on. whatever scale we have used to draw the lengths of our arrows, gives the total voltage across the two components. In addition to this (yes, you guessed it) we find that the angle which this diagonal arrow makes with the horizontal ives the phase of the total voltage across the circuit, with respect to the current! In fact, if we use Pythagoras’ famous theorem about the squares of the lengths of the side of a right angled triangle (whew!), to find the length of this diagonal we find that, if we call the voltage across the resistor Va, and that across the inductor V,, then the total voltage, V, is given by the formula:— VV Looking back to the section on impedance, we notice that this formula bears a remarkable resemblance to the one stated to give the combined impedance of a resistance and reactance; in fact, if we divide both sides of the equation by the current, I, then V becomes Z, Va becomes R and Vi becomes Xi, (since Z, R and X; are all defined to be equal to W/l) and the two equations become one and the same (Howzatt!!) ‘The phase of the voltage can also be calculated, rather than measured directly from the diagram. The appropriate formula is:— n~* Vi/Va, ‘What about parallel circuits? The procedure this time is pretty much the same as for series circuits, but now the ‘reference’ arrow, pointing to the right, represents the total voltage, not the current. The individual arrows represent the currents through the individual components, rather than the voltages, and the diagonal arrow gives the total current, and the angle by which the current leads the voltage. Note that if this angle is multiplied by —1, it then gives the angle by which voltage leads the current. LC Circuits There are two more circuits, that should really be treated by themselves. These are the combination of capacitor and inductor in series or parallel, and they possess some rather interesting properties. If we draw a phase diagram for either of these two types of circuit, we find that the two arrows representing voltages or currents, as the case may be, in the individual components point in exactly opposite directions. To find the arrow that is the combination of these, we place the arrows end to end. That is, we place the of one of them at the head of the other, keeping them pointing in the same directions. An arrow starting at the beginning of the first individual one, and ending where the second arrow does, gives the total voltage, or current. It can be seen from this that if Vo=Vi, then the two will exactly cancel out, and in a series circuit, there will be no voltage across the two components, and the circuit will be effectively shorted across. In a parailel circuit, there will be no current flowing, and the total impedance of the circuit will be effectively infinite, Under what circumstances, then ‘we may ask, will the two voltage (or current) arrows be of equal length, and cancel? It turns out that this is 30 if Xc=Xz, and using the formulae for the reactances of the @) a i Fig. 6a A series LC creuit, where current ls the reference. (b) Phase flngram for the series L-C circu Inductor and capacitor voltages are 180° fut of phase. (2) A parallel LC circuit. (2) With v reference, Inductor and capacitor currents are 180" out of phase- @)‘components, from the section on inipedance, we find that 2nfL must equal 1/2xfC. Here we notice that for any named ‘combination of values for L and C, it should be possible to find some frequency — the so called resonant frequen for which the circuits should react in the way described above. Manipulating the equations, we come up with the formula:— /2xVEE ‘Thus, in a series circuit, signals at this, and only this, frequency, will be able to pass through the circuit unim- peded, whereas in a parallel circuit, any other frequency will be allowed to pass. These circuits are called, respecti vely, a notch filter and a tuned circuit. The latter is of great use in radio receivers, where it is often used to short all signals at frequencies other than those wanted to earth. thus effectively sorting out wanted signals to be amplified and listened to. The frequency required may be selected by adjusting one or other of the two components, and, in fact, the capacitor in the tuned circuit of a radio is usually a variable type, and forms the tuning control. ig. 7. In clreults with more than Gro arrows forthe Individual components can ‘il ie the total voltage or current the voliage or current thea the laal arrow ‘Two’s Company... Of course, you may want to find voltages or currents in circuits with more than two components, but this isn’t as difficult as you might think. Just find the individual arrows of the separate components, and put them all end to end, as in Fig. 7. The final arrow, giving the total voltage or cur- rent, starts at the beginning of the first and ends where the last of the separate arrows does.-Blectronics tod INTERNATIONAL fa ca UNUM Cee cd SUBSCRIPTION ORDER FORM Cut out and SEND TO: Electronics Today International, 513, LONDON ROAD, THORNTON HEATH, ‘SURREY, ENGLAND. Please commence my personal subscription to Electronics Today Intentional wth he ‘se dy FILL UP YOUR MIND We bring you the widest articles you can buy. We control the cost of living. Subscribe now and laugh at inflation for the next 12 months! SUBSCRIPTION am enclosing my (delete a necessary) RATES (ek a appropriate) £71.95 or rues 5 —_ (made payable to AS. id) ‘OR Debit my AcceslBarcayeard (delete as necessary) 55575 or timer Bertie ranging and most informative publish the most original and ingenious projects anywhere. The only thing we can’t do is Cee chequsfonal Grdcsintemationas Money [ ae ‘Order for TTTTT TT IT Pease we BLOCK CAPITALS adn pos codes. ame eg “emi Signature Date aTHERE ARE MANY natural phenomena which are “caused” by time delays. All acoustic instruments and, in fact, everything in acoustics is time related. It is, therefore, hardly surprising that several manufacturers produce electronic time delay integrated circuits. These are called analogue delay lines or sometimes, ‘bucket brigade delay lines’, as this accurately describes their operation _ ~~ anaes al Fig 1. Bucket Brigade delay lines. Quantum Buckets ‘The device can be thought of as being a series of buckets containing water. (Actually it is a series of capacitors con- taining charge). The signal presented to the input fills up the first bucket to the level of that signal. This occurs on phase I of a controlling clock signal. On the second clock (phase 11), all the odd buckets tip their water into the even buckets. No input sampling occurs on clock phase I. On the next clock phase (phase 1) the input is sampled and all the even buckets tip their water in the odd buckets. In this ‘way a signal propagates down the delay line which repre- sents the input signal as a series of ‘samples’. The buckets are really analogue sample and hold units and the tipping is Chapter 2 ANALOGUE DELAY LINES Don’t do now what you can put off for a few milliseconds. Need to delay a signal? Here’s how to do it, plus suggested applications for analogue delay lines. done with electronic switches. This technique is a cross between analogue and digital processes. The charge stored (which is proportional to the input voitage) is truly ana- logue, but it is quantised into small units of time and so, in that sense, it is digital. If the delay line is, say, 512 stages long and the clock frequency is 512 Hz, then the delay time will be number of stages Ixclock frequency? 5° That is, after 0.5 sec a waveform representing the input signal of 0.5 sec earlier will appear at the output. In the example shown in Fig. 1, this signal would only appear at the output for the duration of clock phase II, To fill in the ie | Distr ous — fi Fig. 2. Block diagram of typical delay line system. gaps, a second delay line connected in parallel with the first, but clocked in antiphase, is used, so that a delayed ‘output signal appears on both clock phases Delay lines would seem to solve a myriad of electronic problems but with every solution comes a host of new problems. First, the maximum bandwidth of the delayed signal is proportional to the clock frequency. As the signal is sampled, then the ‘sampling theorem’ says that the signal bandwidth’ must be less than half the sampling frequency, which, for practical purposes, means about one-third. So, if you want to delay an audio signal of 10 kHz bandwidth by 1 second, then the number of stages delay needed is 60,00. This will cost you a few hundred pounds in delay lines. Ifyou choose a lower clock frequency requiring fewer delay lines then you will have to make do on a reduced band- width. If this bandwidth is not controlled by use of an external lowpass filter, then a phenomenon called aliasing ‘occurs which makes the delayed signal sound as if it has been ‘ring modulated’. A typical delay line structure is shown in Fig. 2. A lowpass filter is used to band limit the input signal which prevents the aliasing effects. A second filter is used to recover the quantised output from the delay line by rejecting all the unwanted high harmonics. at “ee line circult basea on the TDA 1022. ‘The input signal level is always larger than that of the output signal because the buckets are leaky, although the leaks occur in both positive and negative directions. Also, the slower the clock frequency the longer the leakage time is and so the loss is greater. This is a major noise generating mechanism. The noise is broad band, being strong in low frequencies (just the area you are listening to), and be- comes louder and more bassy as the clock frequency is reduced. This results in signal to noise ratios of about 70 4B for maximum frequencies. To overcome the poor perform- ance at low frequencies, a noise reduction system such as a compander can be used. The distortion caused by delay lines is typically about 1% and the overload characteristics are not at all good. Heavy overloads can cause the delay lines to stop producing any output at all. The solution is to limit the input level, with some simple sort of diode limiting. One other gremlin is that the output DC level varies with clock frequency which causes some awkward break-through effects. However, once you are fully aware mel al foe TE — | z ‘3. A delay Une ciruit based i onthe SAD S12 of the limitations of delay fines, it is possible to design a wide range of interesting devices. Delay lines work surpris- ingly well when you consider that they move a very small packet of charge through several hundred memory stages with a corruption of only one part in 10,000 to 100,000! egatt Ww ia Fig. 4. A standard CMOS relaxation oscillator. Some Delay Line Circuits ‘Two delay line circuits are shown in Fig. 3. The top one uses a delay line made by Mullard/Signeties. A two phase clock is needed. A preset adjusts the input DC bias so that when the device is overloaded, the clipping is symmetrical. ‘A balance control on the output balances the two outputs for a minimum clock breakthrough. This preset is particu- larly useful when long delay times with audible clock fre~ quencies are used. The second delay line is the SADS12D made by Reticon. This device has the same two preset controls but only requires a single clock signal. There is a complementary clock generator (a divide by two flip-flop) on the actual IC. The input clock must therefore be twice the calculated frequency. ay of Fig. 4b, The frequency ofthis clock generators determined by C1. If long delay times are needed then there is the Reticon S101 which will give you a I second delay at about 500 Hz bandwidth. This device gives a superb automatic double tracking effect (50 mS at 10 kHz bandwidth) but unfortuna- tely it's rather expensive. Clock Generators A selection of clock generator circuits is given in Fig. 4. Circuit A is a standard CMOS relaxation oscillator. The IC costs only about 30p and generates complementary square~ waves; the minimum frequency of operation is about 1 MHz (with a suitable timing capacitor) and the manualcontrol range is about 50 to 1. It is not very practical to voltage control the frequency ofthis oscillator. Circuit B uses an NES66 which is a voltage controllable oscillator IC. The frequency may be controlled via the capacitor Ci, or by interposing a potentiometer or a controlled current source at point X. The output square wave needs to be level shifted and this is done with QI. The maximum frequency using this circuit should be limited to about 100 kHz. For higher operation up to 1 MHz, a faster level shifter is needed. Circuit C uses a CMOS Schmitt trigger and a couple of transistors. This oscillator can readily be controlled by a current generator. The output waveform is a short positive going pulse. A divide by two flip-flop converts this pulse into a pair of complementary squarewaves. Gireuits D and E employ the fast slew rate (13 V/uS) of the Texas B1 FET op-amp range. This enables them to oscillate at high frequencies and to generate squarewaves, with fast edges. Circuit D is a manual control device and circuit Eis voltage controllable. DIY Design A‘DIY" lowpass filter chart is shown in Fig. 5. This filter is 4 4th order Butterworth design. The roll-off slope is 24 t t SLOW ATTACK AND DELAY wy | Fig, 2. A ight dependent restr can be ato mal an analogs sich 7 Gone ~ve, t ig. 4. N-type J FET characterises.A simple electronic analogue switch can be made from a light dependent resistor (Fig. 2). When light from the LED shines on to the LDR, the resistance drops from a few megohms to a few hundred ohms. The device generates very little distortion and noise and is thus suitable for rout- ing audio signals. “The response time of the LDR is very slow, sometimes taking 100 mS to switch, it is both relatively expensive and large and it consumes a large LED current. The transistor chopper, (Fig. 3), is often used as a simple device to turn ‘OFF an analogue signal. This is done by turning ON the transistor and driving it hard into saturation, which in turn shorts the signal to ground. The transistor chopper is often ‘used in organ and electronic piano circuits to produce audi tones by chopping the envelope waveforms with square- ‘waves generated by a tone divider network. It is not easy to use the bipolar transistor as a floating switch, as most applications require but the JFET lends itself readily to this purpose. The characteristics of a JFET (Fig. 4) show that in the region where the curves pass through the origin the device is behaving very much like a voltage controlled resistor. ‘When Vos is 0 V, the curve passing through the origin is, at its steepest, the FET having at this point its lowest Drain Source resistance (Rps), known as Ro. This is typically a few hundred ohms. However, as Vos is increased nega- ar tively for an N type FET, the slope resistance increases until the pinch off voltage is exceeded. The FET is then ‘PINCHED OFF" and the effective Rps is typically a few. | hundred megohms. The JFET can be used as a variable resistance device changing from a few hundred ohms to a few hundred megohms in times much shorter than a micr second. This, combined with smallness of size, low co low power consumption, a very high impedance at its control input and the ability of the drain and source to float, makes the JFET an ideal analogue switch. ‘The series switch (Fig. 5) passes the signal with very litle attenuation when the FET is on and stops the signal when the FET is off. There is a small stray capacitance (a few picofarads), between the drain and source and this causes | high frequency breakthrough when the switch is off. j ‘The shunt mode shorts the signal to ground when the | FET is tumed on (switch off). The attenuation in this mode is R/Roy which requires large values of R to get a large | attenuation. To get the advantages of both types of ar- | rangement, a series/shunt switch can be constructed using two FETs (Fig. 6). When a FET switch is rapidly turned on there is often a click generated, partly as capacitive breakthrough from the gate and partly as an abrupt amplitude modulation of the analogue signal. This can be greatly reduced by band limiting the gate voltage (C2, R3) and by removing any DC _ || ‘component from the analogue signal (Fig. 7). eur oF So orr ig. 7. A slent series switch for audio applications. > w—! Li ov ‘The diode ring switch can operate at very fast speeds.ea Struct = 4-bit ‘multiplying DAC. om mm ig. 100. The ow pase fter frequency response, Ring An Ode The diode ring switch (Fig. 8) is very simple in concept and can be operated at very fast speeds. When A and B are both at 0 V no signal can pass through the ring switch because there is always a reverse biased diode in the route. When A is +ve and B —ve (both with same amplitude), then the 4 diodes are all conducting. Any voltage at the input of this bridge will cause the same voltage to appear at the output. Switch In Packs tis now possible to buy a wide variety of integrated circuits that perform analogue switch functions. There are perhaps over a hundred different types available, many of which are JFET configurations. These are usually rather expensive ind I shall only discuss the more common (and relatively inexpensive) CMOS devices. The CMOS analogue trans- mission gate (Fig. 9) is a bidirectional switch element, having a relatively low on resistance of a few hundred ohms and ahigh off resistance. Gain Code The 4016 is a very versatile electronic building block. It can be used to construct a simple multiplying DAC (Fig. 9). The gain of the amplifier is linearly proportional to the size of the 4 bit digital code and can be programmed to give any one of sixteen gains. A low pass filter (Fig. 10) can have its break frequency programmed by switching in timing capaci- tors. When all the gates are OFF, the timing capacitors are In. The low pass filter can then be tuned by the dual gang pot over the range 1.5 kHz to 15 kHz. When gates B and D are on, the timing capacitor is 11n and so the frequency range is 150 Hz to 1.5 kHz. Also, when A and Care on, the timing capacitor is 101n and so the filter range is 15 Hz to 150 Hz. It is possible to build quite elaborate filter structures, fourth or even sixth order designs and to make them programmable with analogue switches. The analogue switch can be used to modify the parame- ters of an integrator (Fig. 11). The integrator slew rate can be programmed by a combination of RA,B,C and the out- put may be reset to zero using gate D, which shorts out the timing capacitor. = SES Fig. 100, A switched-range low pass filter.ms mo Z Fig 12, Sample and Hold cet, Hold On To It Sample and Hold circuits also employ analogue switches. During the sample period, the storage capacitor is connected to the input voltage via the switch which is tumed on. When the switch is tumed off (Hold period) the voltage on the capacitor has got three possible discharge routes: 1. As leakage across the printed circuit board. This ‘can be greatly reduced by provision of a guard-ring that ‘surrounds the capacitor with a potential exactly the same as that on the capacitor. Surface leakage which is a func tion of potential difference is thus reduced by removing, the potential difference. 2. The switch can discharge the capacitor with leakage currents. These are generally very small and insignificant. 3. The input bias current of the op-amp can discharge the capacitor, but if a FET op-amp is used the effect can be insignifcant. More and Mode The 4051 is an 8 way multiplexer (Fig. 13). This device has a 3 bit binary code input which it decodes internally to turn ‘on one of the eight analogue switches. The 4051 is effecti- vely an 8-way single pole, code selected switch. There is also an inhibit input; a logic 1 at this pin turns off all the switches. po zal Fig. 11 A programmable integrator with rect to er0. ao co mio 13, A ‘tidpier aenatece gone] ofom ) mut choc } ad ae we 20% 4 ao = DOWN —o EOUNTER. Co tt cs) Ca} ig. 14, An audio volume control with 3B steps.wt fe “HH He 0 paraour Per Aro resomceson|oae one Pm! ho t-te | to There are some problems to be encountered when using, this device. The internal decoding exhibits some race condi- tons when the input address changes which may cause switches to momentarily overlap in their ON states. If the 4051 is used a demultiplexer with sample and hold ‘outputs, then the data on each output capacitor gets spread around to other outputs during the overlap periods. This ‘make before break’ effect can cuase serious problems to occur. (There are JFET ‘break before make’ multiplexers but these tend to be rather expensive) Louder Yet programmable volume control can be constructed with a 4051 and an up down counter (Fig. 14). The state of the count at any point in time determines which resistor is in the feedback loop of the op-amp. The resistors are ar- ranged such that there is a 3 dB change in gain for each LSB change in the address code. By making the counter count up the gain will increase and vice versa. It is necessary to stop the counter at 000 and 111 otherwise the gain will roll round giving an abrupt change in level. Fig. 17 (ef). Analogue switches can ano betied Ina aleroproceasor data itrbution system. It is possible to construct an 8-bit analogue Random ‘Access Memory using a 4051 (Fig. 15). Analogue informa- tion is written into memory by first selecting the address and then enabling the ‘write’ switch. The capacitor at this address is charged up to the voltage level of the data in signal. Reading is performed continuously using a FET op- amp so as not to discharge the storage capacitors. This gives the RAM a tristate output capability, which when used with the INH input allows several RAMs to be built into a larger memory system. MPUs Get Theirs | Analogue switches are used in microprocessor data acquis tion systems (Fig. 16). An input multiplexer, addressed by the microprocessor selects which channel to look at. A sample and hold then freezes the selected input signal so that the ADC can perform a conversion on it. National Semiconductors make a single chip that performs most of this data acquisition. The chip, the ADCO817 has a 16 channel multiplexer, an ADC and output latch, but you have to provide your own sample and hold, Analoguevof > FH, t.,. {Th switches are also used in microprocessor data distribution systems (Fig. 17). Data generated by the microprocessor is converted into an analogue voltage by the DAC and is then multiplexed into sample and hold circuits. This is a common technique for systems that require a large number of anal- ‘ogue parameters such as microprocessor-controlled music and speech synthesisers. Plexing Info In most telecommunication systems, a major cost is that of the wire that links transmitter to receiver. If several chan- nels of information can be transmitted down one wire then the cost of the overall system may be considerably reduced (Fig. 18), Eight audio, channels, band limited to 10 kHz in this, example, are fed into an 8-way analogue multiplexer. Each channel is sampled every 50 nS for a duration of 6.25 nS. ‘The eight audio signals can thus be fed down one wire pair as a sequential series of pulses that represent the instant values of the signals in the eight channels. The process is known as Pulse Amplitude Modulation (PAM). wr fia HY fe omer }—_? OF sway | oye satseeatat ceiver cl IRS oo HaHa of} of Tpaggussion 3 otis mein 00 eel Ee pe Fi. 18, The mise and yen ear. Re. sean TEE of jer Fig. 19, (ef) The levels of any numberof miter channels can be on one ‘screen with tls malipexed system, i) Decoding is performed by stripping off the sync signal |) and using it to address the demultiplexer. The required || channel can then be selected by monitoring the respective demultiplexer output with a sample and hold followed by a_ | 10 kHz lowpass filter. It is important that the sampling | frequency is at least twice the audio bandwidth so that | aliasing effects can be avoided. i Mixing It In Monitoring the signal levels in an audio mixer can cause problems, due to the large number that have to be ob- Served. There are now several devices that use a conven- tional oscilloscope or TV to display all the information on fone screen (Fig. 19). Each audio channel is processed by a peak detecting envelope follower. This generates a signal |) that describes the peak signal amplitude in that channel and_|j | it is known as a PPM (Peak Programme) response. These PPM signals are then multiplexed and passed through @ single log compressor to give a display of level in dBs versus channel number on an XY screen. a Tim OrrPuzzled about electronics? HE has the solution! Each month we take the mystery out of electron- ics with regular in-depth features, pages of up-to- date news, and challenging projects for all levels of skill. The key to enjoying your hobby thoroughly is reading Hobby Electronics — every month! WE DELIVERY! Make 1982 the year you start to take Hobby Electronics regularly . . . delivered fresh every month. r= H SUBSCRIPTION ORDER icici et intenatonl ney 1 FORM ler for. mode poyable to A.S.P: Ld) Cut out and SEND TO : ; Hobby Electronics Subscriptions Debit my Accent Barclaycard 513, LONDON ROAD, = THORNTON HEATH, ] T if 1 sc SS re es pre SUBSCRIPTION pe TS for 12 issues RATES 1 tos, sama = eae et an eeChapter 8 BATTERIES In this article we give enough information on the many different types of battery available to allow the right one to be selected for any particular application. BATTERIES may be divided into two general classes; primary batteries and secondary batteries. Primary bat- teries or cells (strictly speaking, a battery is a group of cells, ‘connected together, but the term battery is commonly used for either form), only have a single working life. In general, once discharged their capacity to provide useful power ceases and they must be discarded. A primary cell can provide power as soon as it is assembled and requires no initial charging current. Five types of primary cell are currently available, These are — Leclanche (or carbon-zinc) cell, Mercury cell, Alka~ line cell, Zinc-Air cell and the Weston Cadmium cell. ‘Secondary batteries or cells require an initial charging current before they can be used, in the opposite polarity (oF direction) to their discharge current. They can go through many charge-discharge cycles throughout their useful life, and can be stored for considerable periods in a discharged condition without deteriorating. Secondary batteries are also referred to as storage batteties. Two types of secondary battery are in common use ~ Lead-Acid battery and the Nickel-Cadmium battery (familiarly called the Ni- cad). ‘The Leclanche Cell The construction and composition of a cylindrical Le- clanche cell is as follows. The cell’s outer casing is formed from zine and this acts as the negative electrode. A central carbon rod, which is connected to an external metal cap, forms the positive electrode. The carbon rod is surrounded by a mixture of manganese dioxide and powdered carbon in ‘a porous sack. This is called the depolariser. The rest of the cellis filled with paste of ammonium chloride ~ the electro- lye. The Leclanche or carbon-zinc cell, s commonly known asa dry cell. “These ceils have a no-load terminal voltage of 1VS to 1V6. The energy that they can supply is related to theis size. Under-loading the terminal voltage of dry cells gradually decreases and internal resistance rises. When the load is removed, terminal voltage rises again but not to the original value. Over a number of discharge-rest periods, the no-load terminal voltage will gradually decrease as will the amp- hour capacity of the cell (Fig. 1). Once the no-load voltage drops to 1 volt or so the cell has come to the end of its useful life and should be given a decent burial. Leclanche cells are best suited to applications that require intermittent use oF low-drain use for long periods. Heavy-duty dry cells are available that wil provide much higher discharge currents. These will supply several ‘hundred milliamps for four to five hours at a time, whereas the ordinary cell will typically provide 100 mA or less for similar periods. As might be expected, they cost more than ordinary dry cells. The heavy-duty dry cell deteriorates more slowly than conventional cells and will undergo more discharge-rest cycles before requiring replacement. These are sometimes marketed as ‘Longer-Life’ batteries. Con- ventional and heavy duty dry cells are compared in Fig. 1. Size is not the only factor which governs the life of a dry cell. The ratio of the period of use to the rest period is an important factor. The old door-bell batteries which were about the size of a drink can would last for years. They could supply up to one amp but their rest-to-use ratio was very high. Temperature also affects the performance of dry cells. Optimum is between 20°C and 27°C. Terminal voltage ‘and capacity is drastically reduced below 15°C, and almost ‘useless below 5°C. Leclanche cells deteriorate when stored for long periods. Generally, the larger the cell, the less the deterioration, Leclanche cells have a serious drawback. When left for long periods in a discharged condition, the other zinc con- tainer is gradually eaten away by the electrolyte which then finds its way to the outside, corroding surrounding equip- ment. Leak resistant versions are available, but these should not be left too long discharged either. ‘The internal resistance of a Leclanche cel rises steeply as discharges. This can give rise to low frequency instability (‘motorboating’) in amplifiers. A large value electrolyte (1000 u+) across the supply rail will often eliminate this problem, and will often dramatically improve the sound’ Teproduction from a transistor radio. For applications requiring six volts or more at low cur- rents, the layer battery has been developed. These are made up of square or rectangular layer cells, their shape enabling them to be grouped together with minimum waste of space. The common 9 V transistor radio battery is of type. They suffer less deterioration than the round style dry cell. They are relatively low current devices. Round cells are better where fairly heavy consumption for fair periods is required. ‘The Mercury Cell The mercury cell was invented in World War Il by Dr Samual Ruben. It has an anode of high purity amalgamated Zine and a cathode of compressed mercuric oxide-graphite separated from the anode by an ion-permeable barrier. The cathode is in contact with a steel container which provide the terminal connection. The electrolyte is a solution of alkaline hydroxide, the ions of which act as carriers for the cell's chemical action, The electrolyte is not consumed during discharge. The cell containers are nickel plated steelga Mercury 1.35-1.4 30 3 Alkaline 30 s Leclanche 1:5 6-12 Z12 ie i = Exo onoinany ony Ea os: 1 1 1 Retormanes of rin and heady ry cle charg or sevecl hours per ays fly Ren eure. re ote Dove cat Top mercury eal (Could type SO2R): Cures show — a a ar eran a S$ st ccromne l adapter Steve noes Can Mercurie Oxide Mix -Abvorbant Sleeve nd ‘inane Ciceayie Zine Anode Neoprene Disc Fig. 4. Graph shows discharge performance of slae-carbon, alkaline manganese and mercury batteries, ‘and this do not corrode. Mercury cells are produced in a variety of forms, one of the most common is shown in Figure 2, The no-load terminal voltage of mercury batteries is 1V35. This drops about OV0S to OV1 under load, but unlike Lechanche cells remains very steady throughout the greater part ofits life. At the end ofits life the terminal voltage falls away with increasing rapidity to less than 1V0. Typical discharge curves are shown in Fig. 3. Mercury batteries may be stored for up to three years with only slight deterioration in terminal voltage and amp- hour capacity, Maximum storage temperature is 30°C, opti- ‘mum is 21°C. The characteristics of these batteries are little affected by extremes in temperature. They work well down to ~20°C and some at temperatures in excess of 100°C. Mercury cells are capable of discharge rates much greater than equivalent Leclanche cells, the internal resistance being ‘maintained until near the end of their working life. Mercury cells do not leak if left for long periods. However, their price is several times that of equivalent sized Leclanche cells. A range of voltages is available; typically 1V35, 25, 4 V and 8 V, in different sized packages. A new cell may be used a a voltage reference with an accuracy of OV02 or better. Mercury batteries should always be used within the recom- mended discharge rate for which they are intended — they cannot be recharged. Mercury batteries are used where voltage stability and long life are required. Their small size and high capacity are also advantageous in some applica- tions. The Alkaline Cell 20 10 HOUR DISCHARGE RATE zu ‘NOUR DISCHARGE # RATE TERMINAL VOLTAGE 7 2 é Fig. 11. Charging characteristics of Nckel-Cadanium cells. TRICKLE oCHARGE 7 ame. {obas: %b re (HObRs) Fig, 12. Constant-current charger for Nicad Batteries its normal rate and then immediately recharged. Lead-acid batteries should not be overcharged at high current as this, ‘causes the plates to buckle and slake (which may result in a short circuit). Neither should they be left in a discharged state as the lead sulphide produced during discharge may undergo a generally ireversible physical change resulting in reduced battery capacity. Batteries in this condition are referred to as sulphated. This condition may sometimes be remedied, at least partially, by trickle charging for a consid- ‘erable period. Eventually, sulphated cells self-discharge. In normal operation, lead-acid batteries should be overcharged from time to time, at about half the normal rate, until half-hourly readings of the terminal voltage and electrolyte specific gravity show no further increase. This faction removes sulphate and restores the plates to their ‘normal condition. Spilled electrolyte should be neutralized with an alka.ine solution. This is simply made by dissolving 46 tablespoons of common baking soda (sodium bicarbo- nate) per litre of water, using as much water as necessary. ‘When applied to spilled electrolyte, foaming occurs. When the foaming has stopped the residue should be washed away ‘with clean water. If washing down the top of a battery with. this solution, do not let any into the cells! The Nickel-Cadmium Cell (Nicad) Nicad cells use a potassium hydroxide electrolyte. In a typical unit the positive and negative plates are both perfo- rated steel. The positive plate is filled with nickel hydrox- a ide, the negative plate with finely divided cadmium mixed with a litte iron to prevent it flaking and losing porosity. The electrolyte has a specific gravity of 1.15-1.2, depending on the type of service, it does not undergo any chemical change during discharge. Very little clectrolyte is needed and the positive and negative plates are very closely spaced. ‘Nicad batteries are made in a wide variety of sizes and amp-hour capacities; miniature ones for use in cameras, calculators, etc.,up to large heavy-duty types similar to car batteries. They may be operated over a wide temperature range — similar to that of lead-acid batteries. At low tem- peratures, the amp-hour capacity does not diminish as much as with lead-acid batteries. However, the electrolyte may freeze. As Nicad batteries may be sealed, they can be used in any position. The no-load terminal voltage of a nickel-cadmium cell is typically 1V3-1V4. This drops to about 1V2 volts when discharged. As the electrolyte does not change during discharge (as it does in lead-acid bat- teries), the number of amp-hours obtained from a Nicad battery is much less affected by the discharge rate than are lead-acid batteries (Fig. 10). ‘As Nicad batteries can be made quite small, and can be recharged, they are eminently suitable for use in portable electronic equipment such as calculators, tape recorders, hand-held transceivers, camera flash units, etc. They can withstand considerable vibration, are free from sulphating of similar problems, and can be left in any state of charge without ill effect. Charging should be done with a constant-current charger. The charging rate for the quickest charge should be no more than 1.5 times the 10-hour discharge rate. Most manufacturers recommend a charge rate and a trickle or float charge rate and this is best adhered to. ‘Charging characteristics are shown in Fig. 11. ‘One method of producing a constant current charger is to place a resistor in series with a supply having a voltage three or four times the battery voltage. A better method is shown in Fig. 12. Junction FETs are selected on test for similar dss currents and a number are connected in parallel as shown to supply the rated charge current. The FETs are in series with the rectifier output and the drain-source charac- teristics provide a constant current output. The maximum output voltage should be limited by a zener diode to about 1, 02 aszrs, as2t MOUNTED ON HEATSING 1.2 times the rated battery voltage. “Another constant-current charger is shown in Fig. 13. This will charge batteries up to 12 V at currents up to 10A with a suitably-rated transformer and rectifiers. The output current is selected by closing one or more switches. Lower values of R,, with values according to the formula given, will provide increased output current. . Nickel-cadmium batteries should never be short cir- cuited. This causes internal overheating and the battery may explode, Never dispose of Nicad batteries in a fire or incinerator. ‘This too will cause them to explode! The nickel-iron battery is an earlier counterpart of the Nicad and has similar char- acteristics.Chapter 9 Gm REVISITED (Nothing to do with American car manufacturers) Gm is in fact a throwback from the days of valves, now finding a new lease on life with up-to-date semiconductor devices. MANY A LONG YEAR AGO, when transistors were an item which hadn't been dreamt of by science fiction writers, wwe all used valves, and we all knew the magic letters Gm. ‘Gm stood for a quantity called mutual conductance, and it ‘measured an important feature of the valve from which we could work out how much voltage gain we could get out of a given bottle. Well, the years have passed, and valves are dead for many purposes, but Gm lives and is back working for us. It's odd that Gm should have gone out of fashion for so Jong, because the idea of Gm is even more useful in tran- sistor amplifier circuits than it ever was in valve circuits. Stil, the idea seems to be coming back in a big way, s0 let's take a look at ‘Mutual conductance of any electronic device means the ratio of signal current at the output to signal voltage at the input. For a transistor, this is the ratio Ic/Ve. Ic being the collector current and Vye the voltage between base and emitter, Fig 1. The squiggle above the letters ‘means that it's AC signal voltage and currents we're talking about not the steady bias voltages and currents. Using Gm therefore allows us to represent a valve or transistor as a generator of signal currents, the amount of, ‘signal current being Gm Vin. Now a current generator means a device which will deliver its current into any load, high or low. No valve or semiconductor is really like this, but for most of the uses we make of transistors, the idea of a current generator is not far from the mark. Current Generators If a transistor were a perfect current generator, it would have an infinite resistance at its output. That means just that a signal voltage applied between the collector and the emitter would cause no collector signal current. Once again, it’s not quite correct but not far from the truth. A bit of collector signal current does flow, but not very much, ig. 1. Mutual conductance ieVbe for = about as much as would flow if there were a resistor of around 40k between collector and emitter. Fig. 2. An equivalent ‘reali fora transitor. ‘Now the usefulness of all this is that it allows us to draw an equivalent circuit for a transistor. An equivalent circuit is a circuit made of simple components which behaves in just the same sort of way as some device which is, in reality, much more complicated. A simple equivalent circuit for a transistor is, therefore, as shown in Fig 2. It consists of a current generator, which generates a signal current GmVbe, and a resistor of about 40k in parallel. This simple circuit accounts for the size of the signal current at the output (the collector) and the output resistance between collector and emitter. How does this help us? Quite a lot if we remember all the time that equivalent circuits are about signal currents, not about bias currents. As far as signal currents are concerned, the positive supply line of an amplifier is just as earthed as the earth line. Why? Because in the power supply there's a ‘smoothing capacitor of several thousand microfarads, con- nected between the +ve and —ve lines. As far as DC is concerned, this capacitor is an insulator; but for AC signals the capacitor is just a short circuit, shorting the +ve line to the —ve line. When we connect a ioad resistor between the collector terminal of a transistor and the positive line, then, as far as signals are concerned the load resistor is connected between collector and emitter. Draw this into the equiva- lent circuit, and the result is Fig. 3. Back in the old days of valves (nostalgia corner, this!) we found the sum of these ‘two resistors in parallel, which was Ree_ RL Ree+RL and then the voltage signal out was just the current signal times this resistance (Ohm's Law still rules, OK?) giving Gm Ree RL Ree+RLFig. 3. For AC signals, load resistor connected between collector and positive supply behaves as if connected between collector and emiter- ¢= CHARGE CARRIED BY AN ELECTRON JOLTZMANN’S CONSTANT ‘T= TEMPERATURE IN KELVIN SCALE Je = STEADY (BIAS) COLLECTOR CURRENT Simple Silicon One of the things that makes life simpler in these days of silicon transistors is that the quantity Rce, the output resis tance of the transistor, is quite a large value compared to most of the load resistors we use. An output resistance (the ‘usual symbol is hge) of 40k is quite a bit larger than the 3k3 ‘or s0 we use as a load, so that most of the signal current from the transistor is through this resistor in the equivalent circuit. That simplifies the output voltage to GmRy, so that, the gain of a transistor amplifier is just GmRy. If it's as easy as that, why don't we see it in text books? The reasons are historical — we didn't start with silicon transistors, and a transistor, unlike a valve, doesn’t have a constant value of Gm. If we plot a graph of collector cur- rent against base voltage (as in Fig. 1), the result is not the nice straight line we get when we plot such a graph for a valve, or the not-too-crooked line we get when we plot the graph for an FET, but a very curved line indeed. This indicates that the value of Gm is not constant, but a value which changes as the current through the transistor changes. This, coupled with the rather low output resis- tance of the early gemanium transistors, seemed to seal the fate of Gm for good. Ebers Moll ‘A few years back, though, the Ebers-Moll equation was noticed. If you've never heard of it, you're not alone; very few text books mention it, and some mention it without explaining it. Very briefly, it’s an equation which links the collector current with the Vbe value for a transistor. In other words, it’s the equation for finding Gm. Now the full equation is a fearsome looking thing, full of mathematical symbols you may never have seen before. It repays close attention, though, because most of the symbols are of quan- tities that are pretty well constant, and only two of them vary very much. One of them is the steady bias current, Ic, and the other is temperature. As it happens, temperature, for the purposes of the Ebers-Moll equation, is measured in 40 Vie RL. = 40Vioed Vbe GAIN = 40viosd Vioad 1S THE STEADY (BIAS) VOLTAGE ACROSS AL ig. 4. Transistor circuit wit load resistor (RL). ‘Gan can be replaced by 40k. the Kelvin scale, which starts at the absolute zero of tem- perature around — 273°C. Room temperature is therefore around 293K (no degrees sign) in the Kelvin scale, and a few degrees above or below doesn't make much difference to the equation ‘That leaves Ic as the one thing that really affects Gm, and, the relationship works out at approximately Gm=40le (Iein mA) Put in words, that means we can take a Gm value of 40 times the steady bias collector current in milliamps. For a bias current of 1 mA, the Gm value of a transistor is 40, MA/A. Looks too good to be true, but it really does apply to any silicon transistor, apart from a few freak types. ‘This brings back the Gm idea in a big way, and we can forget a lot of the old formulae we once used in calculating the design of transistor amplifiers. The fact that Gm is not, constant but varies with the bias current is, oddly enough, a help rather than a hindrance. Gain Going back to our equivalent circuit, and ignoring the large ‘output resistance of the transistor, we can now write 40 Tein place of Gm (fig 4). This makes the gain of a transistor with load resistor RL become 40 Ie RL. But Ic in this equation is the steady bias collector current, and so IRL must be the steady DC voltage across RL, the load resistor. This makes calculating the gain of transistor amplifiers with resistive loads a bit easier than falling off a log. Pick a value of voltage across the load resistor, multiply by 40, and that’s ‘your value of gain! For example, we very often design voltage amplifiers so that about half of the supply voltage is dropped across the load resistor. For a9 V supply, that’s 4V5. Do this, and you can expect a voltage gain of 40x4.5=180 times. Don't believe it? It works allright, and tests on a single transistor amplifier confirm it as a rule of thumb. You don't, of course, expect to get a gain of exactly 180 in the case illustrated — there are 20% tolerances on load resistors apart from anything else, but you're never far out; that’s what a rule of thumb is for.When you couple a single transistor amplifier to another stage, of course, that’s another story. You may have set the gain of the first stage to 180 times, but not all of its output signal ends up usefully at the input of the next stage. Reason? The next stage has a rather low input resistance, and feeding signal from the collector of one transistor into the base of another, even if they are directly connected, is rather like feeding signal through a voltage divider. There are, in fact, two ways of calculating how much of the signal is passed on. One simple way is to imagine a voltage divider (Fig 5) in which the load resistance of the first stage forms the upper resistor and the input resistance hie of the second, stage. The quantity hy/Gm, where hy is the current gain of the transistor, a quantity which does vary between one transistor and another. For a transistor with hie=100, Gm setto.40(1 mA collector current) hig is 100/40=2kS. If we feed this from a transistor with a 4k7 load resistor, the amount of signal reaching the second transistor is 25 T3447 = 35 of the signal at the output of the first. This brings the gain of the first transistor stage down to 180x.35=63 which is the sort of value we usually measure for one stage of a multi- stage amplifier. With all this going for it, Gm is coming back, folks. As ‘Sam Goldwyn is supposed to have said, “simplicate and add lightness”. Let's hope we've added a bit of lightness. K. T. WilsonChapter 10 PL L’Lis being used for mass production of LSI ICs, but little has been said about the theory behind this new technology. This article explains how it works. INTEGRATED INJECTION LOGIC or PL can provide serious competition to CMOS circuits where minimum cur- rent and high component packing density is required. De- vices using PL circuitry can be produced very economically and the speed of operation rivals that of TTL. This new technology is being used by some of the major semiconduc- tor manufacturers for products ranging from microproces- sors to quartz-controlled electronic watch devices. All PL devices are large scale integration LSI products — they ‘contain a very large number of components on a single silicon chip. PL was developed quite separately (in Europe) by Philips and IBM around 1972. It employs bipolar devices, (that is, devices like conventional transistors rather than FETs) in circuits which have been derived from the early DCTL (Direct Coupled Transistor Logic). It is only quite recently that developments in the I°L production processes have made this circuit technique economically attractive. ‘A DCTL circuit is shown in Fig. 1. Three transistors are shown in each of the three NOR gates with the output of Gate 1 feeding one of the inputs of both gates 2 and 3. Other connections, which are not shown, are made to the other inputs of the gates. Circuits of this type were used in simple $SI (Small Scale’ Integration) devices, but suffered from the disadvantage that the current was unequally divided among the transistors in any one gate owing to minor differences in their base-emitter voltages. In addi- tion, the load resistor had to be separated from the trans- Mig. i. A Direct Coupled Transistor Loge circuit DCTL). siderable area of the chip. ‘Note that in the circuit of Fig. 1 there are direct connec- tions between corresponding regions of the transistors: all of the emitters are joined together, whilst the two bases which are driven from the collectors of gate 1 are common. ‘The current to these bases passes through the load resistor of the gate 1 circuit. In an IL circuit, these common electrodes share the same area on the chip. A cross section through an FL gate is shown in Fig. 2 and the circuit is shown in Fig. 3. A single pnp transistor is employed as a current source to supply current to many transistor bases without the use of a load resistor. The whole of the emitter region is a common one beneath the surface structure on the chip. This eliminates the need for surface metallisation for each separate ground connection. In addition, the area required per transistor is greatly reduced. IBM initially used the name Merged Transistor Logic (MTL instead of It should be noted’ that the pnp transistor is formed laterally along the surface of the silicon chip. The other component is a muli-collector apa transistor characteristic of FL devices. However, this npn transistor is formed vertically in the silicon. The n-type epitaxial layer acts as the grounded emitter of the npn transistor and also as the ‘grounded base of the lateral pnp device. The p-type base of the multi-collector transistor also serves as the collector of the pnp device. Thus the two devices do not exist as separate structures.Injection np transistor ‘injects’ current into the base of the multi-collector transistor — hence the name Integrated Injection Logic. Current from a current source (not shown in Fig. 3) passes to the emitter of the pnp transistor and hence to the collector. Switching of the logic state occurs when this current is switched to or from the base of the multi-collector transistor. ig: Cron section through an FL gate, — Be RANEY. Ve) If the input at the base of the multi-collector transistor is ow (less than about +0V7), this potential will be inade- ‘quate to overcome the natural forward junction potential of the npn base-emitter junction and the npn device will be non-conducting. The injected current will flow out of the input connection to the collector of the previous circuit (not shown in Fig. 3). The multi-collector transistor outputs will therefore rise to the ‘high’ logic level, this voltage being determined by the collector circuitry. If the input voltage now becomes ‘high’ (that is, over +0V75), the npn transistor will be biased to saturation and the output of the collector will be ‘low’. This low value can ‘be about 0V02. Thus the change of the logic level is repre- sented by a voltage swing of around 0V7. Power Supply ‘The positive power supply line of PL circuits is connected only to the emitters of the pnp injection transistors. The base of these transistors is earthed, so the I°L circuit as viewed from the power supply line is effectively just a forward-biased silicon diode. The total power supply cur- rentis therefore the sum of the currents fed to the injection transistor emitters. ‘The voltage levels in 1°L circuits can be very low; indeed, such circuits can operate from a supply of 0V8S upwards. ‘The supply current per gate can be very low (about 1 nA), but the injected current can be increased in value up to about 1 mA to permit switching of the circuit at a much higher speed. ‘Although the FL circuits can operate at_ such low voltages, the input and output circuits normally included in the same package require a higher supply voltage and their requirements normally determine the operating voltage of the whole device. A series voltage-dropping resistor is used in the power supply line of some FL devices, whilst other devices incorporate a voltage regulator on the chip to elimi- nate the need for an external resistor. ‘The use of an internal regulator circuit also enables various injector current levels to be obtained at different points in the circuit so that each part can operate at the minimum power level for the switching speed required by that particular part. For example, the fast frequency divi ing circuits of a quartz controlled watch can operate at a high injection current for a satisfactory performance at 32 kHz, whereas the following frequency dividing circuits operating at a low frequency can use lower injection current levels. The increased cost of fabricating such circuits may be well worth while when current consumption must be minimised. In many applications a single dry cell can be an ideal power source for I°L circuitry. ‘A guard ring of n+ material (shown in Fig. 2) is required in PL devices to reduce cross-talk between adjacent gates However, this ring can touch the base of the npn device and itoccupies little surface area. 1 cater Fig. 3. The rel represented bythe gate shown above a Fg 2 Gates PL gates can be made by “wire-ORing” the isolated col- lector outputs as shown in Fig. 4. Similarly NAND gates ccan be made by using the multiple collector outputs of the npn transistor connected as shown in Fig. 5. Input/Output Circuits PL is almost always used in conjunction with other cir- cuitry. The voltage change when an I’L circuit switches is only about OV7 at current levels which may be very low. If the inputs and outputs of the FL circuits were brought out ectly to external connecting pins, any small stray noise pulses or interference picked up by the circuit would be likely to trigger the P-L circuitry, owing to its great senstiv- ity to low amplitude pulses. Buffer interfacing circuits are therefore used between the input and output connections of a device and the [AL cit. cuitry itself. A typical input buffer which can accept TTL. input pulses and convert them into pulses suitable for the ‘operation of an FL circuit is shown in Fig. 6. The input buffer circuit used with some of the older logic systems can, be even simpler. ‘An output buffer circuit which can amplify the low ‘voltage pulses from the output of an °L circuit and provide enough current and voltage to drive a TTL input is shown in Fig. 7 Technology Comparison An E°L gate can be made with what is effectively a single ‘component on a chip area about one tenth of that required for a normal three-component CMOS gate. In addition, IL is one of the most economical technologies used in device fabrication, since the number of masking and diffusionoperations on the silicon slices are less than in most compa- rable techniques. One of the advantages of IPL technology is that it is so very similar to that of other standard linear and Schottky TTL manufacturing processes that it is easy to fabricate other type of component on the same chip. For example, light emitting diode driver circuits can be built on the same chip as PL circuitry; this enables a single chip to be used to drive the display of a watch or a calculator as well as to carry out the required logic operations. Operational ampli- fiers, oscillators, voltage regulators, etc., can be fabricated ‘on chips containing I°L circuitry. Fig. 4 (above) PL gates can be made by “wire-ORing’ the isolated collector output —D=p Fig. 5. (above) NAND gates can be made by using the maltiple The CMOS process is essentially suitable only for the production of purely digital devices, although simple de- vices such as transistors and diodes can be fabricated on the chip. In contrast, Schottky TTL devices can be combined with PL circuits om a single chip to produce produets which are faster and which have higher component densities than can be achieved in other ways. The Texas Instruments ‘SN74S201 and SN748301 356 bit random access memories are examples of such products. ‘The power consumption of IAL circuits increases linearly with the speed of operation required and in practice you can use the minimum injection current required for maxi- mum speed at which the circuits will ever operate. CMOS circuits consume very litle power in the quiescent state, but the power required increases with the switching speed. ‘Thus no circuit adjustments or settings need be made if, ‘minimum power consumption is important and the maxi- ‘mum operating speed is always available. In other words, MOS circus always consume, minimum power at low operating speeds, but have a high speed capability demand” vinereas FL circuits must be adjusted for low power or high speed or some intermediate value of power 4g Fig. 7. An output buffer which can amplify PL pulses to feed teTTL clreatry. Fig. 6. Atypical input buffer which can accept apa’ pe an conver dem for tae With ‘irl. consumption and speed. PL is faster than CMOS, whilst Schottky-clamped FL is, even faster still. The silicon-on-sapphire version of CMOS another way of obtaining faster logic devices, but Emitter Coupled Logic (ECL) offers the highest speed at the ex- pense of ease of use. The susceptibility of PL devices to noise pulses has al- ready been mentioned. CMOS devices require input pulses with an amplitude of about half the supply voltage used and are therefore very resistant to spurious operation by stray noise pulses. It is difficult to see how future I°L can be fabricated without input and output buffer circuitry because of the noise problem. ‘A comparison between the various logic systems is given in Table 1. Applications FL is employed in a wide range of applications which require large scale integration. It is unsuitable for making devices with only a few gates so it seems most unlikely that simple FL logic devices will become available (like those ‘one meets using CMOS and TTL technologies). Brian DanceChapter 11 CMOS 555 APPLICATIONS The result of bipolar versus Fig. 1. The pla-out and internal configuration ofthe CMOS $55. CMOS in the 555 league. PARAMETER IM7555. BIPOLAR s85C ‘QUIESCENT CURRENT TYPICAL "TYPICAL VeewisV Towa ima INPUTCURRENTTRIGGER ‘SOA os A YHRESHOLD SpA out A RESET 100g Ot A MAX. OPERATING FREQUENCY S00 klz ‘00 ke POWER SUPPLY RANGE 2p 18V AvSpI6V PEAK SUPPLY CURRENT toma 370 mA "TRANSIENT RISE AND FALL TIME 400s 100 as ‘ATOUTPUT Fig. 2. A comparizon between the Bipolar and CMOS version of the S55. Fig. 3. The CMOS S55 displays an impressive reduction in sappy current {ranslent dering an output tranamisslon- THE BIPOLAR 555 TIMER CHIP has been around for many years, but there is also a CMOS version that has some very significant design improvements. The two devices are functionally very similar, being interchangeable in most applications. The operation of the 555 is very simple (Fig. 1). It con- sists of a pair of comparators that operate at ¥s and of the supply voltage, this being set up by a resistor chain. These comparators set and reset a flip-flop which in turn drives the output stage. A second output is available which is an electronic switch (Discharge) to ground. Other features include access to the resistor chain via the control voltage pin and an extra reset input to the flip-flop. This simple network readily lends itself to all sorts of oscillators and timer circuits. ‘The bipolar 555 has a few parameters that can make it difficult to use, but whichhave been improved in the CMOS version (Fig. 2). The bipolar quiescent supply cur- rent is generally about 10 mA which negates their use in small battery units. The CMOS version consumes a mere 120uA. ‘Also the CMOS inputs are very high impedance having input currents down in the pico amp region. Another major improvement is the reduction in the power supply current transient during an output transmission. The bipolar is very noisy in this respect and can often be the cause of lots of ‘funnies’ in nearby circuits. The CMOS 555 is a low power, high input impedance, device that should be used where low current consumption is at a premium. The following circuits illustrate some pos- sible uses of the device:oy ras ect moc a per esc Rice oo uesnetteannm iene Serra nase aaa Se here maedt ama ‘tecarge ET (pa) tarot on. Ths docharges te capacitor ia Rb Seerecie naam eae reas “aia eeescaee re ick cr wu acer hao hetecenieenciaeens BEE Raison eet ee coe at “Eker peet ee cereent aetna secret aeact erence ae auc keoeemaan aaa Sea ee pa a mata JA fo generates suwtonth waveform by reducing Rb to his eases the rest time to be Yer fast, ofthe order of «few Fi, 5. (above) This oscillator has cal output because the charge ‘and diocharge paths are the same andthe portion ofthe eure ‘hat i used fs symmetrical, Note that inthis creat, the dacharge pin (7) is svailble to do other Jobs, suchas drive x LED or some other device. The timing resistor R shouldbe kept relatively high (above 10) to preveat Toading ofthe output. Sn a 50 ON period is produced by D1 (which s forward biased) and R2, C1. erage current consumption is relatively low. If the unt mere powered, froma ¥ battery, then the current would be 120 uA for the ICM 7855 and [an average of 140 GA forthe LED, making 260 0A total. This would give = IMfetime of few months fora PP3 which would be extended by the OFF period (increase Ri) and reducing the LED cucrent increase R3) short th A Fig. 7. (above The 7555s wed toate and terminate a rggered sweep, einai dhcharge FEY (on De ONand Clu horde road, ‘When trgger to theft, the collector of 2 me Meee serine hw OF cone, Olt Ra ana Dh {ema current generator that drives Cl. Once the discharge FET has bee firmed OFF, the volag on Ci iss ineary- When tis aage reaches 2 ‘Verte threshold comparator state discharge FET info Ws ON sate and 20 ishored to ground: 1C1 inane to buffer the voltage on Cl. The Sirep generator lot retriggerable and sony tated om fst postive ung put, Fo vary the rwecp rate, iter C1 tnd or 2. h #. {sore A DC 19 DC converter can econ fam an cater adhe charge forms a high frequency square wave ‘sclator, The aguarewave from pn 3s AC coupled ia C2 to the charge ‘Pump, The vligeon the negative side of C2 prevented from going more {han OV postive by DI andloo te usrewaneon this sie of he capator biases cso that l mores from + OV to -8V3 Dz charges up C3 onthe ‘egnve excorson ofthis waveform ads a negative fall of about “8 Vs fencrated. The current that canbe taken rom this al rater low, beng ‘termined bythe oslbation frequency and C2. General DC to DC ‘omverters haves poor tanaer eicleacy.aw wn Sebete ig. 9. (above) The 7585 can be used asthe driving oscillator in an ultrasonic remote control system. The scllator generals a thin pulse seut Su ong athe natura resonant reueney ofthe transtce. This is used torn on a transistor (Q1) which drives an auto ‘witha 10to | step up rai. The output ofthe transformer fx gael ranadacer and when tion frequency correc 100'V peak to peak sinewave will be at ls point The transducer Sul crwth high impedance and soa high operating voltage ‘equlred to produce any power output, The reelver fa 40 kis bandpass ‘iter. Ths will amply any audlo signals a hls frequency, which can then besent toa detector circuit. Ei ig, 10, (above) This oscilator allows the mark space ratioatpin Stabe OF ‘arid rom | to 20, to 20t0 1, by wang two feedback igh, Cl ls charged via DI, part of BVI and tL, When iis tow, Cl charged via DZ, the other part of RVI and ‘ucllation frequency i lightly dependent on this rato. it impedance triangle ‘put, the frequency of which maybe controled with he IM log pot ig. 1, (abo) By mang og oat 756 an ete nd an "2 trlangle/squarewave excilator is produced, To operate well at (ap to 40 Lil) a CMOS inverter should be used to replace Inverter. The op-amp provides alow = ‘rr conn sare ig, 12. (above) The 7555 can be used to generate an acoustic tone, The ‘ostllator ise to rum at TK3 Hz which has alow period of about 1S nS and a period of about 758 nS, Daring the ow period the transistor i turned isthe ouopeaer ia commie acrw the poe supp and ik routes, When pin Sis atabout 100 mA (for a9 V supply). This ives ica ‘he collation, As the transistor i only on for 15 out of every 7 msthe postionof the RVI average current through the speaker i quite small ~ about 1.95 mA. ‘wiper determines the ratio of tue charge and discharge periods, The ‘Therefore, the total current consumed by the whol system is only about 2.5 ‘A a3), and yet the 0 mW output signal i quite andibe,Fig, 13. A warbling tone cam be generated by using two oscillators. The svarble is produced by TCI which generates a 13 He waveform thats used Torrequeney module the fone generator as described inthe previous trample Pin Sofa 7588s connected tothe + Voc lap onthe resistor || tndder1y tying tt awarble” waveform, frequency modulation ofthe ||| Smal output tone is produced. A 7886 could be used insted of two 7555s. Tosco aL Fig. 16 (above) The 7555 can be made into a monostable although some Desblene may occur ints ue. A negative going pulse on the trigger imput {hin can be used to start the monostable period Its important that this alse gocs high again before the end of the monesiable period, or ese it may ‘Prolong the perf. To this end an AC coupled transistor inverter bas been ‘sed so that rising postive signal wil initiate the event, Titally C215 “discharged to ground, When pin 2is taken low, the discharge FET is st to SeOFF and so the volinge at Eres witha te constant of C2, 3, When this voltage reaches +24 Vee, the discharge FET ts turned ON, C2is, ‘ fe ground and the monostabie pried I flaished. During this fered te 725 produces high outpt at in. Fig, 14. (above) The police and other Tis trequeny bee on tei red setworka This doc’ inerfere mh the normal radio trafic und allows the listener tobe certain that bef stil |l}. tuned tm cocrecty to that channel. The clrult generates a similar ‘beep’ fand yet consumes only a couple of zliamps. TCI sa show aslo second period) with large marklspace rato, “The dlcharge FET is on for most of the time and only gues OFF for sbout || 1Stlliseconds in every 3 seconds. This FET is connected tothe fone ienerator In such a manner that when the FET is ON the fone generator i Ethbited. When the FET goes OFF the generator produces a burst of 3 KHz ‘onclatons which are heard as a beep" he Fig. 17. (above) The previous creult has been modified, The timing resistor fhe been replaced witha programmed current source, IC2, Whatever {current is put imo pin Sof the CA 3080 (the Tape current) wil appear atts Tie wl ipearly charge up C2 when the FET switch is turned OFF. ‘Th monostable action vil be the sare as nthe previous example. The ‘omostable period is nearly proportional tothe Iagc current 90 by Programming ths current the period is controled: 7 encase rom von ason@nn aunt Fig. 15. (lef) The last ype of sound, tobe described isa simple firtn, IC generates asaraooth waveform which i used to frequency ‘modulate, via pn 5, the fone generator (1CZ). As the sawtooth vollage rises {lth a period of 8) so does the tone generator frequency. ee agdetermined byte age caren Bes ep igh tipo ure coanet the dining a ste Tacit ha exelent hgh aoe wf TT : ee ea oe LeU ORE then FLO (200 0A quiescent could be se. Pech one thar rte newtonth wna andChapter 12 TV PRINCIPLES Ever wondered just how your TV actually works — all those cunningly interconnected and interrelated bits of high-voltage circuitry? Stay tuned for enlightenment. ig 1. Circuit diagram of UHF varicap tuner, which uses two RF amplifiers and a mixer stage. Resonant lines tuned by Capacitor diodes selec the required channel and provide the necessary front-end selectivity. THIS ARTICLE looks at monochrome television based on ‘a mains/battery chassis from Thorn Consumer Electronics In addition to producing an insight into modern television technology, the article also gives a fair impression of how the picture is developed on the screen of the picture tube. ‘The basic principles are common to all receivers except that for the reception of colour there are circuit additions for the decoding of the colour information (and a tricolour tube for display!) ‘Sound and vision signals are modulated on to two carrier waves, the former using frequency-modulation (FM) and the latter, amplitude-modulation (AM). On the prevailing UK 625-line system the signals are transmitted in Bands IV and V which are located in the UHF spectrum. Each chan- nel occupies a width of 8 MHz with the sound carrier being 6 MHz above the vision carrier. For example, Channel 21 hhas sound and vision frequencies of 477.25 and 471.25 MHz respectively, while the frequencies for Channel 68 are 853.25 and 847.25 MHz. Vision modulation is negative-going (See Fig. 3), and is transmitted in 5.5 MHz upper sideband and 1.25 MHz lower sideboard. Peak FM deviation is 50 kHz (as distinct from 75 kHz on FM sound radio) based on 50 uS pre- emphasis. Ratio of peak vision to peak sound power is 5.1 Further information on the signal is given later. Tuner-Front-End The start of any television receiver is the ‘front-end’ or tuner (Fig 1), whose job itis to select the required channel and to convert the sound and vision carriers to lower fre~ quency ones for subsequent intermediate-frequency (IF) amplification and response tailoring. In Fig. 1, the aerial signals are coupled to an aperiodic RF amplifier, VT1, through an ‘isolator’ for preventing spurious mains voltages in the receiver from reaching the aerial at lethal power! The transistor is in common-base mode so that the input is applied to the low-impedance emitter. Further input match- ing is provided by the emitter components and the base is biased either from a constant potential or from an AGC potential (sec later). ‘VT2 is a tuned RF amplifier, also in common-base mode, but the tuning is by resonant lines rather than coils. Any transmission line whose length is adjusted to correspond toras used for G and M a tuned frequency is the equivalent of a tuned LC circuit. ‘An open-circuit line is resonant at 1/2, 3/2, 5/2, etc. wave- length. Excluding velocity factor, the physical length of a line for, say, Channel 33 would be around 280 mm. Hap- pily, it is possible to reduce the physical length while retain- ing the required electrical length by cutting off the ends of the line and replacing them with capacitance, which reduces the physical length to about 50 mm. Moreover, tuning the channels then become a question of varying the capacitance atone end, Looking at line L6 in Fig. 1 shows that the bottom connects to varicap W1 and the top to C8 and VT! collector capacitance. A varicap is essentially a junction diode. As the reverse bias is increased so the depletion region widens, and as this constitutes the dielectric between the n and p regions. The effect is tantamount to the two plates of a capacitor being moved away from each other, with a conse- quent reduction in capacitance. The four varicaps in Fig. 1 are biased by a positive potential being applied to the “tuning volts’ input. The potential is obtained from a stabi- lised (by IC1) supply in the main chassis (Fig. 2) via tuning potentiometer R39. Thus as this control is tuned so the resonance frequency of the lines alter in step and tune over the UHF channels. The second RF amplifier stage VT? starts to give selectiv- ity. Emitter coupling is via low impedance aperiodic line LT. Further selectivity is provided by the bandpass coupling between VT2 collector and VT3 emitter formed by lines L1I, L12, L16 and L14, tuned by varicaps W2 and W3. Common-base VT3 uses collector/emitter feedback for the local oscillator tuned by line L17 and varicap W4. Line L15 couples the RF signal to the oscillator/mixer stage. The circuits are trimmed by LS, L10, L13 and L18 (so they all tune in step), while the closed lines L3, L8 and L21 also assist with the tuning and matching The oscillator is arranged to operate at the IF frequency above the input frequency, and additive mixing yields the IF output, which is resonated by L23 and associated compo- nents. The IF signal is coupled to the IF input of Fig. 2 via C30. ‘The high degree of selectivity minimises spurious responses such as image, IF, repeat spot, etc., while also providing a good 3rd-order intermodulation rejection ratio. This is further aided by the nature of the transistor and design of the first stage VT 1 The circuit also reveals various signal coupling, decou- pling and isolating components, which are essential for the stable performance of this important part of the receiver. The tuner is built into a fully screened box with feed- through capacitors for the inputs and outputs. IF Channel Sound and vision signals of the selected channel undergo amplification with band pass and selectivity tailoring in the IF channel comprising VT1/2/3/4. Tuner signal is applied to VTI base from the tuned coupling L1/C3, and the amplified and bandpass defined output is yielded by transformer Lialb. Gain is controlled automatically (AGC) to suit the level of the input signal by a bias fed to the bases of VTI and VT2. The four stages are each in common-emitter mode, and impedance matching at the couplings is achieved essentially by capacitor divide-down. The bandpass characteristic is provided in the main by LYCI/C3/C1OR1 at the input and by a L7a/b at the output. Additional selectivity is provided by collector inductor L2/L3/L4, while sound and adjacent channel sound rejec- tions are introduced at 33.5 MHz by LS/C18/C19 MHz and at 41.5 MHz by L6/C21/C22. With the 625 line system (system ‘I’ is used in the UK) the sound carrier is 6 MHz above the vision carrier, but because the local oscillator of the tuner is working at IF above the signal frequencies, the IF appears at 33.5 MHz for sound, which is 6 MHz below the 39.5 MHz vision IF. ‘The sound and vision signals are handled simultaneously by the IF channel, which is possible because frequency mod- ulation (FM) is used for the sound signal. Vision bandwidth of the ‘T’ system is 5.5 MHz upper sideband, accommodated by the IF bandwidth, and overall channel width 8 MEV.Vision Detector Sound and vision signals from L7b are coupled to vision detector W1, which yields a changing amplitude output corresponding to the picture information (Fig. 3) and also an output at 6 MHz resulting from intermodulation of the sound and vision signals by the diode nonlinearity, the difference frequency of the two signals being 6 MHz. The intercarrier sound signal (as it is called) retains the FM of als because this is one of the components If the ratio of the levels of the sound and vision signals is incorrect a buzz occurs on sound — called intercarrier buzz. Hence the reason for the 33.5 MHz trap, which sets the sound signal level below that of the vision carrier while helping to establish one side of the bandpass. The 41.5 MHz trap avoids the sound signal from the next channel causing interference while helping to establish the other side of the bandpass. The vision carrier is set 6 dB down the response to equalise for the sigle side band signal. Video Channel Picture and intercarrier signals from WI are directly coupled to the base of the video driver VTS via low-pass filter L8/C27/C28, which removes residual IF signal. VTS collector is loaded into transformer L10 which tunes the 6 ‘MHz intercarrier signal and couples it to the sound section for FM demodulation and subsequent pre and power amp! fication for driving the loudspeaker. VTS also serves as an emitter-follower for the video signal with network R26/R27/R32 as the load. The signal‘across this is directly coupled to the base’ of the video output transistor V7, which feeds negative-going picture signals to the cathode of the picture tube from its collector. A series rejector L9/C46 tuned to 6 MHz is also active at VT7 base to prevent intercarrier signal from getting into the video output stage, where it would cause picture interfer- cence. The level of video signal reaching the tube is adjusta- ble by R47, the contrast control, which is a kind of current feed-back control working by the progressive shunting effect across R47 emitter resistor by R48. C49 is a DC isolator. Video-frequency compensation is also provided by capacitors in the feedback loop. VTS base is biased from a resistive divider complex (R24/R39/R2etc.) from the supply rail. It is also partly biased from rectified IF signal at WI anode, and since direct-coupling is used an increase in IF signal level results, in a reduction in positive bias at VTS base and hence a fall in potential across VTS emitter load, ‘The voltage across R32 (the preset contract control part of the load) is fed to the base of the AGC amplifier VTS at a level established by the setting of the control. Because VT6 collector is energised via W3 from positive-going 5 pp pulses derived from a winding on the line output transformer (bottom right-hand corner of Fig. 2, next to the picture tube), the transistor conducts only during the line syne pulses when there is no picture content which the AGC circuit might otherwise falsely read. The degree of conduction and hence the level of the collector potential are determined by the DC level of the line sync pulses at VT6 base. This is called line-gated AGC. il i tl‘Thus with increase in input signal level (such as when tuning to a strong channel) VTS is turned down and the positive potential at its collector rises. This is reflected via forward conducting W 10 to the bases of VTI and VT2 by way of R2/R6 and the filter consisting of C34/C35/C36/R35, which removes line pulses. The small-signal transistors VT and VT2 are the type designed for forward AGC; that is, increased gain reduction resulting from positive-going AGC potential, ‘The preset contract control R32 sets the operating range of the AGC. With a test card signal of average strength the control is adjusted for 1.5 Vpp picture plus syne signal at VTS base. Some sets include delayed AGC for the tuner RF ampli- jer which comes into effect after the gain has been reduced initially on a strong signal by the IF AGC; but for the ‘monochrome portable this is barely necessary as maximum front-end gain is generally necessary for most of the time for the best signal-to-noise ratio when a simple set-top acrial is utilised. It will be seen that the tuner ‘block’ in Fig. has an AGC input which, in this model, is terminated to a supply potential-divider. Field Timebase The electron beam needs to be deflected both vertically and horizontally to build up the raster upon which the picture appears. The vertical deflection is handled by the field timebase which deflects the beam from the top to the bot- tom (scanning stroke) and then very swiftly back to the top again (retrace) at 50 Hz repetition rate. ‘This is achieved by a 50 Hz sawtooth current passing through the field scan coils (L15) on the tube neck. The ‘oscillatory requirements are provided by the field oscillator VTI8/19, which is an RC multivibrator. The retrace is initiated by the arrival of a field sync pulse at VTI8 base (Gee later), while the repetition rate is determined by the vertical hold control R116 with R117/118 and C102. Consider the circuit during the scanning stroke when a rising voltage (ramp) occurs at the base of high-gain ampli- fier V120 owing to C104 charging through R127/128. This turns on VT20, VT22 and VT24, and turns of VT21 and VT23. At the conclusion of the stroke VT24 is fully “bot- tomed’, at which time a positive-going pulse from VT19 collector ‘hits’ the bases of VI21 and VT22 via the multivi- brator isolating diode W19. The pulse is initiated from the field syne action. The retrace is thus triggered by VT21 and ‘VT23 tuning on, and VT22 and VT24 turning off. During the retrace, VI24 collector voltage rises at a rate established by the L/R ratio of the scan coils, and when the supply line voltage is exceeded W21 goes into reverse con- dition and VT23 is isolated. The rate of rise is then defined by C109. After a peak, the retrace voltage falls until W21 ‘goes into forward conduction again. This allows the re- ‘mainder of the retrace energy to be fed back into the supply line, after which the scanning stroke recommences. The resulting rise in current through the field scan coils during the scanning stroke produces a magnetic field such that the electron beam is drawn downwards. To avoid vertical non-linearity to the display the rate of change of current must be linear. Owing to resistive losses in the scan coils and circuit non-linearities, a slight correction to the TR OS AG — cee seanme| "AtEP | *aurs| acetone =e current waveform is required, and this is achived by @ parabola waveform produced by R138/R137/C106 being added to the ramp via the linearity amplifier VT20. The degree of correction is adjustable by the vertical linearity control R137. When the retrace is initiated, the rapid reversal of scan coil current deflects the beam swiftly upwards to start anew downward scan, and during the retrace diode W20 goes hard into forward conduction so that the base of VT20 is clamped to earth.Line Timebase Horizontal deflection of the beam is achieved by the line timebase driving a sawtooth current wave through the line sean coils (L14) during the scanning stroke. Deflection is from left to right, and at the end of the scanning stroke a swift reversal of current deflects the beam back 10 its starting point again. During the retrace a considerable amount of energy stored in the inductive elements of the line output stage is released to provide the extra high ten- sion (EHT) for the final anode and the high voltage for the first anode (A1) of the picture tube. Boosted voltage is also used to energise the line output transistor VT17 once the line oscillator has started. Line repetition rate of the 625 line system is 15625 Hz. ‘Thus the horizontal rate is significantly greater than the vertical rate. We have seen that in the UK the vertical rate is 50 Hz. This means, then, that a raster of 312% lines is produced (15 625/50). For a complete picture there are two vertical scans, each producing a raster of 312% lines, so that the complete picture is made up of 615 lines and produced every twenty-fifth of a second (in actual fact not all the lines are used for the picture as some occur during the field sync period when the electron beam is cut off). ‘A complete full-line picture is achieved because the scan- ig lines of one field interlace in the spaces between the lines of the partnering field. To obtain 625 lines without interlacing the line frequency would need to be increased to 31 250 Hz. This in turn would call for a greater rate of change in beam intensity and hence spot brightness to trace out the fine detail over each line, and because a greater rate of change or signal amplitude involves a greater bandwidth, more radio space would be needed to accommodate the picture detail of each channel. With the 5.5 MHz vision bandwidth of the ‘I’ system good definition is obtained at the 15 625 Hz line rate Interlacing could be avoided without using up extra radio space by reducing the field rate to 25 Hz, but then the picture would suffer bad flicker (subjectively apparent up to about 45 Hz). Interlacing thus solves the problems of bandwidth and flicker without unduly detracting from the displayed information. Returning to the circuit in Fig. 2, the line frequency is ied by a blocking oscillator incorporating VT1S forward base bias through R83 turns the transistor on so that the current through the collector winding of L12 rises. The reversed phase of the other winding puts a negative: going pulse on the base which cuts the transistor off. The ‘on/off cycles are timed by L12/C83/C84 with the oscillator ‘on the free-running mode, the frequency being set by L12 core. The oscillator is synchronised to the line pulses of the signal (as will be explained later). ‘VT16 amplifies and shapes the pluses from VT15 emitter and transformer T3 couples them to the base of output transistor VTI7. The pulses switch this transistor on during the scan so that current flows through the upper left-hand windings of the line output transformer (LOT) TI and scan coils LI4. Because the coils are essentially inductive the current rises as a fairly linear ramp. However, because the effective length of the beam changes with scanning stroke owing to the wide scanning-angle and flat screen of a contemporary tube, ‘S-correction’ is required. This is ‘achieved by C93 which reduces the rate scan at the start and’ end of a line with respect to the centre. Further linearity correction and width adjustment are provided by a closed- loop sleeve set under the scan coils. The field produced by the current induced into this counteracts the non-lineatity of the field produced by the scan coils themselves. VTIT switches off at the end of a scan and the swiftly collapsing current through the scan coils and LOT windings returns the beam to its starting point and yields a high voltage pulse owing to the sudden release of the induc- tively-stored energy. The repetitive pulses are increased in voltage by the overwind at the top right-hand side of T1, rectified by W14 and smoothed by a capacitance formed by the inner and outer conductive layers on the tube flare, the inner connected to the final anode. The result is a potential of 11.5k V for the final anode. After rectification by WIS, pulses from the lower right-hand winding charge C95 to yield a 95 V line for the tube first and third anodes, video ‘output VT7 (to provide about 50 V video swing for the tube) and varicap tuning Oscillatory energy is rectified by the booster diode W12 conducting during the retrace to charge C87/C88. This not only damps the unwanted energy which would otherwise cause vertical lines at the left of the picture, but the poten- tial developed from it is used to energise VT7 collector, and contribute to the line scan, thereby improving the efficiency of the line output stage. The stage also adopts 3rd harmonic tuning of the pulses. This tends to flatten the tops of the pulses, which leads to improved EHT regulation. ‘The tuning capacitor is C89 in parallel with a low-induetion disc capacitor C92 providing flashover protection. Syne Stages Video signal at VTS emitter is coupled to the base of the syne separator transistor VT 14 through R72/C72. On the 625 line system the picture"signal is negative-going (mod- ulation level falling with increasing brightness), and at the end of each line sync pulse occurs whose tip reaches 100% amplitude, as shown in Fig. 3a. Composite video (picture plus sync) from VTS emitter is fed to the base of VI14 (syne separator) which is biased to conduct only during the syne pulses so that line syne, coupl- ing is to VTIL (syne amplifier/inverter), whose output drives ‘flywheel’ discriminator W6/W7, etc. The discrimina- tor is also fed postive-going line pulses from the LOT via (C94 which, after RC integration, form a ramp whose phase is compared with that of the line sync pulses. Phase error results in a potential at the top of R78 which, after being filtered by C78/C79/R82 to remove pulse residual is applied to the line oscillator. As this is a VCO, frequency correc- tion and hence line synchronisation are achieved. From the end of one field scan to the start of the next one, five narrow equalising pulses are followed by five broad field sync pulses and then by another five equalising pulses. The width and spacing of the pulses keep the line synchronised during the field syne period, while the equalis-ing pulses ensure equal blanking on both even and odd fields, and also identify the two fields for accurate interlac- ing by cutting off the picture half way through a line at the end of odd fields and starting it after a line is half over at the beginning of even fields, as shown in Fig, 3 (b). It is worth noticing that test signals and certain teletext data are transmitted on blank lines ~ the latter at a bit rate ‘of 7 megabits per second. The 1.55 and 5.8 us front and back porches to the line sync pulses provide time for the line retrace, and it is the 5.8 us porch which carries colour burst signal. “The positive-going field sync pulses at VT14 collector are integrated by C99/R124 and applied to VT18 base through WIS. The integration builds up a composite pulse for trig- ering the field retrace and attenuates line pulses. Tube Biasing and Video Feed During normal working the tube grid is held at chassis potential by W17. When the set is switched off WI7 is everse-biased and the charge held by C96 drives the grid negative, thereby suppressing the beam, while the supply voltages collapse. Beaming current cut-off is set by R105 (brightness control) which merely adjusts the tube cathode potential. Video signal from VT7 collector is also applied to the cathode, and as the signal is negative-going the beam cur- rent increases with increasing picture brightness. Beam cut- off or black-level is set by the brightness control so that the syne pulses drive the tube below black Sound Channel Intercarrier signal from L10 is fed to IC2 which incorpo- rates a 6'MHz limiting amplifier; quadrature coincidence detector tuned by L11; voltage-controlled attenuator ‘operated by the volume control R54 and an audio preampli- fier for driving the class B push-pull output transistors VISIVTI2 via driver VT8. The bases of the output tran- sistors are driven together from VTS collector, which is possible because VI9 is NPN and VT12 PNP (a comple- ‘mentary pair). Quiescent current is set by R59 at 8 mA. ‘Negative feedback is from the emitters of the output pair via R57 to VTS base. Since the mains supply is isolated by transformer T2 it is possible to use a headphone set or ‘earphone connected to jack J1. For those not familiar with the quadrature FM detector the following brief description may help. After passing through the limiting amplifier chain, the inter-carrier signal is changed to squarewave and the signal fed two ways: one ‘way to a synchronous detector and the other way to a 90- degree phase shift circuit and thence to the synchronous detector. The synchronous or coincidence detector com- bines the two inputs vectorially so that the output consists of the vector sum which, relative to the fixed 90-degree phase shift, changes with the FM deviation. The result is a variable width squarewave (pulse width modulation) which, after integration, yields the audio signal. Power Supplies ‘The receiver can be operated from a 12 V car battery or the mains supply. On mains, isolation is provided by transfor- mer T2 and full-wave rectification by W8/W9, with C70 the reservoir. The supply is fed to the emitter of series regula- tor VT10. VTI3 is the error amplifier which compares a ratio of the collector output voltage with a reference poten- tial provided by zener WS. Starting current is provided by R66 and the base potentiometer R69 sets the output voltage for the correct value of EHT voltage. Stabilisation is effective over a mains input of 220-264 V. The high Vie rating of VTI0 provides automatic protection against re- versed battery polarity. Finally, one or two minor points: SPI at VT7 collector is a spark gap which liberates energy in the event of a flashover inside the tube, directing current away from VT7 collector. ‘The tube is a quick-heat type whose heater is energised from the 11.3 V line and one which is happy with a relat vely low focus electrode voltage. Gordon King "You will do the following ‘Build s modern oxiloxcope 1 Recognite and handle current electronic components (© ess, draw and understand creult diagrams (© Carryout 40 experiments on basic electroni eeuits used in modern equipment © Build and Use digital electronic fhreuts and curent solid sate ‘chine {© Leain now to test and sevice avery typeof electronic device Used in industry and commerce todsy ‘Servicing of Rad, TV. THF! and mer oproceerorleomputer equipment Newob? New Career ? New Hoty? GetintoBlectranies Now! _ DBs to GE Soon I l POST NOW TO: I BritishNational [Radio&Blectromics School soit _ Bane Ban TBR Chesca “aidde cars ecease |DIGITAL COMPUTERS, folks, are not always fastest or cheapest. No, don’t faint. No kidding — those amazing digital circuits do not always get the jobs done first! They're fine, of course, as long as we stick to straightforward arith- ‘metic, but unfortunately, there are occasions when we want to do other things such as integration or differentiation. Circuits performing these functions are not only of use in computers, however, they are of great use to those of us ‘who are just simple mortals, as well. For instance, in func- tion generators, a squarewave may be changed to a triangu- lar wave merely by integrating. Now, while digital circuits can perform these functions, they do tend to get a bit bulky and expensive. It’s very ‘much easier to use analogue circuits. As it happens, we have very simple networks that make passable integrators and differentiators for very little money. They're capacitor- resistor series circuits, and their operation is quite easy to understand. Differentiators ‘We can make quite a serviceable differentiator circuit from the series combination of resistor and capacitor shown in Fig. 1. Now from our original definition of capacitance: the current flowing through a capacitor is given by ave a But in the case where we are driving a load with very high input impedance, Iour will be negligible, and Ip will be very close to I.. We can say, without too much inaccuracy, that Tr=I.. Ip, however, is given by Ohm’s law, Tn=Vour/R. Thus Vour=Rik. Since Ip is the same as I., however, this gives A Vour=Rk=RCGe and so the output voltage is effectively the voltage across the capacitor, differentiated and then multiplied by a scale factor RC. If we don’t want this scale factor, we can just arrange matters so that RC = 1. ‘The main problem with this circuit, of course, is that itis, indeed, the voltage actoss the capacitor, and not that across the input, which is differentiated. However, as long as we don’t let the output voltage get too large, V will be very close to Vix, and this error will not matter too much. Chapter 13 OP-AMPS AND INTEGRATORS Although we’re dominantly digital these days, there are some applica- tions where the trusty op-amp is cheaper and easier — differentiating and integrating circuits, for instance. Fig 1 Basi iterentiatr. Fig. 2. Basic iterator. Integrators ‘The basic integrator circuit is very similar to that of the differentiator ~ the resistor and capacitor just swap posi tions (Fig 2). Now we can find the circuit's action in the same way as we did before =c4Vour 1.=C- dt Integrating both sides ofthe equation Jledt=CVour. But [is the same as Ip, provided we are driving a load with high enough input impedance. From Ohm's law, we have Ig=Va/R, and thus URJVpat ‘Vout: Dividing both by C Vour=l/RCf Vad. Again, the voltage being integrated is the voltage across only one of the components — the resistor — and not that across the entire circuit. However, as long as we again arrange that Vou, that is, Vc, never gets too large, Vp is, very close to Viv, and we have a fair approximation to an integrating circuit with a gain of URC. Bigger and Better So far, the circuits we have looked at have had two main disadvantages; they are accurate only when driving circuitswhich have very high input impedances, and their output voltages cannot be allowed to become very large, or the difference between the input voltage and the voltage actu- ally being acted on becomes too large to be ignored. (This in turn puts resistrictions on the allowable values of RC time constants and thus the components themselves, but we ‘won't go into that.) How can these problems be solved? Did the man at the back mention op-amps? Dead right, friend. To see how they might be useful, however, let's doa ‘quick bit of revision on them. (Those familiar with op-amps skip the next section.) Op-Amps Op-amps are famed for three major properties. The first of these is a very high input impedance, the second is a very low output impedance and the third is a gain so high that it may be approximated to infinity without too much inaccu- racy for most purposes. It is this last property which leads to the ‘virtual earth’, a very useful concept in analysis of op- amp circuits. ‘The voltage gain of an amplifier is, by definition, the ratio of ts output voltage to its input voliage. If the gain is m, then ‘the output voltage Vour is Vj, or, if we are using ‘the inverting input of an amplifier, —mV,yj. However, as we have stated, the gain of an op-amp is close to infinity. Thus, its output voltage is infinity times its input voltage, or, putting it another way, the input voltage is equal to the ‘output voltage divided by infinity. Since the output voltage must be finite, the input voltage, or, more accurately, the difference in voltage between the inverting and non-invert- ing inputs, of an op-amp, must be zero. (Yes, I know it Jooks as though I've cheated somewhere, but I can assure ‘you that it works.) Since this difference in voltage is zero, it follows that if we ground one input of an op-amp, the other input automatically goes to zero potential. This is not to say that it automatically gets shorted to earth — there is still @ very high resistance between the two points — it just means that no voltage will be present; there is a ‘virtual earth’. ‘This concept, as has been stated, is a very useful one. Now wwe can apply it to our integrator and differentiator circuits. The New Improved We saw in the last section that an op-amp has a very high input impedance and a very low output impedance. It was a very high input impedance, you will remember, that we needed for our basic circuits to drive, so suppose we put ‘some sort of unity gain voltage amplifier on the outputs. It ‘wouldn't affect the signal in any way, but it would mean that we could drive circuits with lower input impedances. ‘Well, using an op-amp, a unity gain voltage amplifier has a circuit something like that shown in Fig. 3. It's easy enough to understand; the output is shorted to the inverting input and the voltage present at each is identical. However, the difference in voltage between the two inputs must be zero and so the same voltage is present at the input to the amplifier as is at the output. In practice, this means that the ‘output voltage follows the input voltage. Amplifiers like this are often used as ‘buffers’ — allowing high output impedance circuits to drive low input impedance ones. If we put one of these buffer amplifiers on the output of each of our circuits, we have the circuits shown in Fig. 4, ‘and we have, indeed, solved one of our major problems; the circuits no longer need to drive into high impedances. ‘The other problems still present, however. Is it possible to improve our circuits again? Well, yes. Fig. 3. Unity gain voltage amplifier, = ig. 4, A differentator with buffer. ig. 4b, Integrator with buffer. . Sa, (above) Integrator circuit using an op-amp, and below) differentiator circuit using an op-amp.AtLast... ‘We'll look at the integrator first. The circuit is shown in Fig. 5a, and does not use an op-amp tacked onto the end, but instead, as an integral part of the circuit. (Yes, that's right, it’s an integrated integrator). It’s action is as follows: Since the input impedance of the op-amp is very high, it follows that the current actually flowing into itis very small, and hence In = ~ Ic, to a first approximation, in order to keep the currents flowing into point A sum to zero. (Kirch- hoff’s first law — the algebraic sum of all the currents flowing into a point of a network is zero. This is the same as saying current in = current out.) However, Ohm’s law tells us that the current flowing through the resistance is given by the voltage across it, divided by the resistance. Now, the voltage at A is zero (virtual earth), so the current through the resistor is Viy/R. ‘The current through the capacitor is given by Your a dt Hence, we have, since current through the resistor equals current through capacitor ViwR four it and so vovnc=—S¥gur Integrating both sides of the equation, we obtain qdt=—Vour oF Vour=—URC§ Vit. Since R and C are constants, and can thus be moved out of the integration sign Hence we have effectively a circuit which integrates input voltage with respect to time, and which has, once again, a gain given by — 1/RC. The integrating action may.be seen if we apply a squarewave to the input, We obtain a triangu- lar wave as output, and one which compares very favourably with that obtained from our original circuit (Fig. 6). ig. 6a. Input squarewave signal; b. output from op-amp circuit «output from orignal circu. Fig. 7a (above) Input squarewave signal, and b. (below) Differentiator Mark 3 fi The action of the differentiator circuit (Fig. Sb) can be | explained similarly. Again, current through the resistor is || equal to that through the capacitor, because of the very | high input impedance of the op-amp. | L=-Ig But Ip is given by (Vour—Va)/R and Vq is zero (virtual | earth again), Similarly, Ics given by CdVpy/dt. Therefore Vout cl¥s ‘multiplying both sides of the equation by R, we get Vour=-R“¥iN And we have a differentiating circuit, the gain of which is | given by SRC. We can see the differentiating action if we | apply a square wave to the circuit asin Fig. 7. utpat from op-amp circu. r Howzat! ! With these two circuits we have overcome the difficulties experienced with our original RC combination series cit cuits. The voltage being acted on is the input voltage actu. |) ally and, thanks to the low output impedance of the op. amp, we can use these circuits to drive many more circuit {H ‘The output voltage, which we were forced to restrict in our original circuits for fear of affecting the action of the cir cuits, is now restricted only by the supply voltage to the op amps ‘The outputs of these circuits are, of course, inverted, asis I! shown by the minus signs in our equations, This is because of practical dificulis incurred when a non-inverting c= cuit is used, and can easily be solved by tacking a unity gain inverting voltage amplifier onto the output ~ surely a small price to pay forall the advantages that these circuits give us bver the originals, A. Lipson |}IG Microcomputing SUBSCRIPTION ORDER Today Subecrtpsione 513, LONDON ROAD, THORNT! Please commence my personel subscription to Computing pees seen eee ey ke ee Fora personal approach to Your contemporary guide to today’s news and tomorrow’s technology. 1am enclosing my (delete Cheque Peetal Sra omy (made payable io ASP: Ley Debt my Accens/Barclaycaré* ‘ice on neces) 1 COED TTT TTT Please ute BLOCK CAPITALS and include post codes. [Name (Me (Mea/Mis) See aeee Address =Chapter 14 EDDY CURRENTS The life story of Eddy Current, last known to be circulating in the region of discs and transformers. THE BRANCH OF PHYSICS now known as electro- ‘magnetism can be said to have been born in 1819. It was in that year that Professor Oersted of the University of Co- penhagen discovered that electricity and magnetism are related — that a current flowing in a conductor produces a magnetic field in the close neighbourhood of the conductor. Later, around the 1830s, the reverse effect — that an ‘electrical current can be produced in a conductor by a changing magnetic field — was discovered simultaneously, and quite independently, by Faraday in England and Henry in America Both of these effects are used, for example, in the trans- former; an alternating current in a coil creates a changing magnetic field, which, in turn, is used to produce an EMF (and hence a current, should a circuit be connected) in another coil. However, rather less people are aware of another, very closely related, and extremely interesting, effect — the phenomenon of eddy currents. What's In A Name Magnetic fields are not usually quite as selective as we would like them to be. A changing magnetic field will not only produce an EMF in any coils in its vicinity, but it will also produce EMFs, and hence currents, in any conductor around = even any old lumps of metal that may be just hanging about. These currents don’t actually go anywhere — they just circulate round and round within the conductors, like eddy currents in a liquid. Hence the name ~ eddy currents. Since eddy currents are the result of induced EMFS in conductors, and because resistances within conductors, can be very small, the current can on occasion be quite sizeable, and so the effects produced by them can be very significant. Jn fact, eddy currents are far more chan just « scientific curiosity. Depending on exactly where they are, and what they are doing, they can be either a curse or a blessing. However you view them, though, they are an interesting phenomenon, and can produce some fascinating effects, not all of which are totaly useless! Count Your Blessings . . . One of the more striking experiments on eddy currents is shown in Fig. 1. A horseshoe magnet is suspended on a thread, above an aluminium disc which is itself free to turn about its centre. If the magnet is now spun round, the aluminium disc starts to rotate with it (although it never Quite catches up with the magnet). Similarly, if you spin the aluminium disc, the magnet above it also starts to turn. This obviously cannot be due to ordinary magnetic effects — THREAD ROTATING HORSESHOE MAGNET FREELY TURNING ALUMINIUM DISK aluminium is non-magnetic, and if you try to pick up the disc with the magnet, you will find that you are unable to. It is apparent that something funny is'going on. (No, air currents aren't dragging the disc round when the magnet rotates — you can put a sheet of paper between the two, and the effect still works!) Field Study The relative movement between the magnet and disc is inducing eddy currents in the aluminium. These, in turn, create other magnetic fields, and it is these that cause the ‘magnet and disc to move together — the magnetic field of the magnet interacting with the fields caused by the eddy currents (sounds a bit like pulling yourself up by your bootstraps, but it's correct). An interesting follow-up to this ‘experiment is to replace the disc with one cut as shown in Fig. 2. The slots tend to get in the way of the eddy currents and prevent them from flowing, so such a disc is notrealy reduced, oF even stopped dragged round so easily by a magnet (which is another way ‘of showing that air currents don’t do the work ~ the slots shouldn't make any difference to them). Interestingly enough, this apparently insignificant effect ‘actually has some practical application, It is used, for in- ‘tance, in the normal car speedometer! The rotation of the wheels is transmitted, by various means, to a magnet, which itself rotates, with a speed proportional to that of the wheels. This rotating magnet induces eddy currents in an ‘aluminium disc (or its equivalent) and tries to drag it round. However, a spring is used to hold the disc, so it is unable to turn very far. The faster the car goes, though, the faster the ‘magnet rotates, the greater the eddy currents, and the further round the aluminium disc is pulled. By attaching @ little red or orange needle to this disc and seeing how far this needle rotates, we can work out how far the disc has turned, and hence the speed of rotation of the magnet Thus, we find out the speed of the car. Yes, I wish I'd thought of it first, too. Cutting Your Losses Besides being useful, though, eddy currents can also be very annoying, They could justly be called the transformer designer's nightmare. The transformer is, basically, two coils close together, However, in the middle there's a dirty igreat lump of metal (the core) and its doesn't just sit there doing nothing, with all those magnetic fields about. No prizes for guessing what happens. It might not seem that ‘eddy currents in the transformer core would be much ofa PURIFIED, coit CRYSTALLINE SAMPLE HIGH-FREQUENCY A.C. problem, but they are, for two main reasons. Firstly, the ‘eddy currents mean a loss of power in the transformer and hence reduced efficiency. It stands to reason that if power is being used to drive currents around in the core, then that much less power is going to be available for use from the secondary coil. The second problem is no less serious, especially in large-scale transformers. The power being wasted in the core, driving eddy currents round, quite naturally ends up as heat, and consequently transformers are liable to get very hot. Indeed, large transformers, such as those on the national grid, may be oil-cooled, to prevent overheating, Itis obvious that, in transformers at least, eddy currents, are not wanted. So what can be done about them? Well, if you've ever taken an old transformer apart for the wire, or even just out of curiosity (naughty, naughty) you will prob- ably have noticed that the core is not just one solid lump; it is built up of flat metal laminations. This is not because they make the cores out of flattened baked bean tins. The lami- nations are separated by varnish or paper or some other insulator and this greatly increases the internal resistance of the cores, reducing eddy currents. Hence, both the loss of power and the unwanted heatings are reduced. Even the heating effect of eddy currents can be put to use, though. It is used in the production of pure crystalline samples of conductors like metals or semiconductors ~ germanium, for example. The impure sample of the material is passed in a crucible, through a coil, which has passing through it @ high frequency alternating current. The magnetic field produced by this current induces eddy cur- rents in the specimen and the heating effect is great enough to melt it! As the sample passes through the coil, the molten zone within it is carried to one end (Fig. 3). Impuriti within the sample are accumulated in the molten zone and hence get taken to one end of the specimen. This end is later removed. What is left is a very pure, crystalline sample of the substance. So eddy currents can be surprisingly use- ful! Footnote ‘There is one final point which must be at least mentioned in connection with eddy currents. This is the induction motor, an indispensible servant of industry. It depends for its ‘operation on eddy currents . . . full explanation of ‘though, is another story altogether. ‘A. Lipson IMPURE SAMPLE MOTION — OF CRUCIBLE Fig. 3. The heating effect of high frequency AC can be pul to good se in semiconductor material manufacture.