November 1987 Rs. 7.50 A Cheap Kit Can Cost Much More! LOW GRADE COMPONENTS.' EXAGGERATED RATINGS AND SPECIFICATIONS. BAD WORKMANSHIP. INADEQUATE INSTRUCTIONS. INFERIOR PCB MATERIAL TRACKS COMING OFF DURING SOLDERING BURNT OUT TRANSFORMERS DUPLICATE COMPONENTS All these factors can add to the final cost of your project, and can cause unbearable harrassment, so, when you buy an Electronic Hobby Kit-insist on: m VEQAKIT K Always the first Choice! r~ Available at: r □ gala 20, Kalpana Building, 1st Floor, 357, Lamington Road, Bombay 400 007. CLCU 1 KUIUIlfS Tel: 363549 LATEST ADITION TO VECJAKIT RANGE • ROULETTE WHEEL • MULTI TUNE CAR REVERSE HORN • LISTENING BUG • SELECTIVE TUNE MUSICAL BELL • AMERICAN SIREN • TDA 2006-10W AUDIO AMPLIFIER • CAR REVERSE HORN (UM356I) • LM 317 T VARIABLE POWER SUPPLY Economy offers complete line of Digital Instruments for measuring Electrical Parameters and problem solving temperature measuring, monitoring and control systems — all designed to improve product quality and cut manufacturing costs through efficient energy utilization. Milli Ohms Meter Digital Thermometer Digital Multimeter Digital Temp. Controller Digital wattmeter Digital Ammeter Digital voltmeter pocket Digital Thermometer POCKET DIGITAL THERMOMETER FOR QUICK RFSPONSE, HIGH ACCURACY Et INTERCHANGEABLE PROBES. ONE FOR EVERY JOB SOLVE ELECTRICAL MEASUREMENT PROBLEMSl WITH RUGGED, DIGITAL IMONITORING INSTRUMENTS! Digital Calibrator Digital Insulation Tester Economy offers most advanc- ed battery operated portable instrument for temperature measurement for industrial and laboratory applications. Compact, accurate & reliable with LED flourescent displays ensure clarity of read out. Temp : Range : — 50°C to 1200 C. Immersion Probes: for molten metal temperatures. Surface Probes: for moving cylinders and rollers' in textile & calendering machines. Fast Surface Probes: for mea- suring temperatures of elec tronic components. Other products: Temperature Controllers, Digital Tempera- ture Indicatiors, Pt. 100. Thermocouples & Compensa- Gas Probes: for Air-condi- t< n 9 Cables etc. tioning & Furnace tempetature measurement Manufactured & Marketed By: Economy ELE is. Sweet Home, ptot no. U2, 2nd Floor PitamDer lane. Off. Tulsi PiDe Road, Manim, Bombay-400 016. Tel.: Office: 2861165.2862360 Gram: ECONHEAT. Telex: 011-3892 ECON-IN. Manufactured & Marketed bv: Economy EiECTRomcs 15, Sweet Home. Plot No. 442. 2nd Floor pitamber Lane Off. Tulsi Pipe Road, Mahim, BomDay-400 016. Tel.: Office: 2861165,2862360. Gram: ECONHEAT Telex: 011-3892 ECON-IN 11.04 elflktor mdia november 1987 Publisher: C.R. Chandarana Editor: Surendra Iyer Technical Editor : Ashok Dongre Circulation: J. Dhas Advertising: B.M. Mehta Production: C.N. Mithagari Address: ELEKTOR ELECTRONICS PVT. LTD. 52, C Proctor Road, Bombay-400 007 INDIA Telex: (011) 76661 ELEK IN Overseas editions: Elektor Electronics 1, Harlequin Avenue, Great West Road, Brentford TW8. SEW U.K EditorLen Seymour Pulitron Publicacoes Tecnicas Ltda Av Ipiranga 1 1 00, 9° andar CEP01040 Sao Paulo — Brazil Editor: Juliano Barsali Elektor sari Route Nationale; Le Seau: B P 53 592270 Bailleul — France Editors: D R S Meyer: G C P Raedersdorf Elektor Vorlag GmbH Susterfeld-StraBe 25 100 Aachen — West Germany Editor: E J A Krempolsauor Elektor EPE Karaiskaki 14 16673 Voul8 — Athens — Greece Editor: E Xanthoulis Elektor B V Peter Treckpoelstraat 2.4 6191 VK Beek — the Netherlands EditoriP E L Kersemakers Ferreira & Bento Lda R.D, Estefania. 32.1° 1000 Lisboa — Portugal Editor: Jorge Goncalves Ingelek S.A. Plaza Republica Ecuador 2-28016 Madrid-Spain Editor: A M Ferrer In Part: Kedhorn Holdings PTY Ltd Cnr Fox Valley Road & Kiogle Street Wahroonga NSW 2076 — Australia Editor: Roger Harrison Electronic Press AB ,Box 63 1 82 1 1 Oanderyd - Sweeden Editor: Bill Cedrum The Circuits are for domestic use only The submission of designs of articles of Elektor India implies permission to the publisher to alter and translate the text and designed and to use the contents in other Elektor publications and activities The publishers cannot guarantee to return any material submitted to them All drawings, photographs, printed circuit boards and articles published in Elektor India are copyright and may not be reproduced or imitated in whole or part without prior written permission of the publishers. Patent protection may exist in repect of circuits, devices, components etc described in this magazine. The publishers do not accept resonsibility for failing to identify such patent or other protection. MEMBER Printed at : Trupti Offset: Bombay - 400 013 Ph. 4923261, 4921354 Copyright © 1987 Elektuur B.V. The Netherlands Electronics Technology High-current switching regulator 1C simplifies supply design 1 1 .24 Electronmicroscopy 11.26 Filters:- theory & practice Part-3 1 1 .30 Switch-mode power supplies 1 1 .45 The positive impedance converter 11 .50 Research collaboration to boost IT 1 1 .52 The desk-top supercomputer 1 1 .53 Background to hollow emitter technology 11 .54 Projects Low-noise microphone amplifier 11 .34 14-bit D-A converter 11.36 SSB adapter 11 .42 Drill speed control 1 1 .44 Information Electronics News 11.20 Telecommunication News 11.22 Computer News 1 1 .23 New Products 11.64 Readers Services 11 .74 Datalek 11.77 Corrections 11.80 Long life bulb Seven segment display Frequency measurements with a multimeter elektor indie november 1987 1 1 .05 HIGH CURRENT SWITCHING REGULATOR IC SIMPLIFIES SUPPLY DESIGN by Giuseppe Gattavari* Containing a complete 2.5A switching regulator on a single chip, a power IC simplifies the design of many power supply schemes and reduces cost. While designers are attracted by the high efficiency of switching regulators, they are often deterred by the com- plexity of circuits based on con- trollers such as the SG3S24 and discrete power transistors. By integrating on a single chip a complete switching regulator capable of delivering 2.SA at SV to 40V, the SGS L4960 High Cur- rent Switching Regulator 1C of- fers all the advantages of a switching regulator yet is little more complex to use than a lin- ear regulator. A 2.5A/5V regu- lator can be built with one L4960 and just eight compo- nents (Fig. 1) and higher output voltages are obtained by ad- ding two resistors. Moreover, the device includes current limiting and thermal protection circuits, eliminating the need to add extra circuitry. A further advantage is that the L4960’s source-sink output stage can switch in about 70ns, allowing efficient operation (up to 90%) at switching fre- quencies of 100kHz. The IC is, in fact, tested dynamically at 100kHz in production. Thanks to the high frequency operation the output LC filter components can be very small. The IC itself is assembled in the compact Heptawatt 7-lead package— both horizontal and vertical mounting versions are available— and requires only a small heatsink. Considering also the few external compo- nents and small LC filter the complete application circuit is extremely compact (see photo). It can even be squeezed into the comer of a system card. A versatile device, the L4960 may be used in a number of dif- ferent ways. The most obvious is a basic DC- DC converter configuration where a 50/60Hz transformer, rectifier bridge and filter ca- pacitor feed the input of the device with an unstabilized DC voltage. SAWTOOTH OSCILLATOR PWM S COMP >SO«jM OUTPUT STAGE 5 IV REFERENCE R t : t r CC'M^ Multiple outputs Any number of devices may be combined to produce a mul- tiple output supply, permitting a building block approach to supply design. In multi-chip supplies it is desirable to synchronize the switching fre- quencies and this can be done by connecting the oscillator pins in common to one RC net- work (Fig. 2). 87176-1 2.2 «F i Fig. 1. Containing a complete 2.5A switching regulator with current limiter and protection functions, the L4960 High Current Switching Regulator needs few external components. 11.24 elektor india november 1987 3 BVV 20-50 fr40W>l 87176-3 Fig. 3. A low-current auxiliary output can be obtained by adding an extra winding on the output inductor. This technique is simple and inexpensive, yet works well provided that power drain on the auxili- ary output does not exceed 20% of the power delivered by the main output. Fig. 4. With two extra windings both positive and negative auxiliary outputs can be produced with just one device. 5 87176-5 Fig. 5. An L4960 can be used as a pre-regulator in supplies where on-card linear post regulator are employed. 2 87176 -2 Fig. 2. So that several devices may be combined to produce modular multi-output supplies, the switching frequencies can be synchronized by connecting the oscillator pins together. Very often low current auxiliary outputs such as + 12V are re- quired. In this case it may be necessary to use several devices since an auxiliary out- put can be produced by adding an extra winding on the output industor as shown in Fig. 3. In this example the secondary winding is not isolated— the bottom end is at 5V. This means that fewer turns are necessary than if it were isolated, and load regulation is improved. Circuits of this type have the ad- vantage of low cost and sim- plicity, and yield satisfactory performance as long as the power drain on the auxiliary output is no more than about 20% of the power delivered by the main output. Extending this principle, positive and negative auxiliary outputs can be obtained by ad- ding two windings, as shown in Fig. 4. Since designers always prefer to avoid inductors when- ever possible it should be noted that all of these circuits require small toroids with a small number of turns. Pre-regulation In all of the applications de- scribed above the L4960 sup- plies the load directly. The device may also be used effec- tively as a pre-regulator for supply schemes where on-card linear post regulators are used (Fig. 5). The advantage of this approach is that it enhances regulation at the expense of ef- ficiency. Using an L4960 as the pre-regulator overcomes the poor efficiency of post regu- lation without making the supply excessively complex. The overall efficiency of this scheme can be further in- creased by using new-gener- ation low drop linear regulators like the SGS L4941 for final regu- lation. These 5V/500mA regu- lators have a maximum dropout voltage of 650mA, allowing the use of a lower intermediate i voltage. Offline switching supplies Another important application is where the L4960 is used to regulate auxiliary outputs in high power supplies where the mains transformer has been re- placed by an offline switching 6 Fig. 6. In a conventional offline switching supply, linear regu- lators are used to stabilize the auxiliary outputs. This solution requires a costly transformer and dissipation is a problem. regulator to reduce size and weight. To understand the advantage of using switching regulator chips in this application it is necessary to examine the drawbacks of the conventional circuit, illustrated schematically in figure 6. In this example, a feedback circuit guarantees the necessary precision for the main 5V output while conven- tional linear regulators stabilize the other outputs. A major drawback of this supply design is the cost of the transformer. A separate sec- ondary is needed for each out- put and each must be opti- mized to obtain a low drop-out voltage, otherwise power dissi- pation wil be excessive. The linear regulators used in these supplies are either ICs (for currents up to 2-3A) or discrete circuits. Either way, power dissipation also be- comes a problem in overload and short-circuit conditions since the current limiters are of the constant-current type. It is therefore necessary to com- plicate the circuit further by ad- ding thermal protection circuits or to use a large, and costly, heatsink. A final consideration concerns the rectifier diodes between the secondary windings (one for each output voltage) and the filter capacitors. With a linear post regulator the input current is always equal to the output elektor india november 1987 1 1.25 current so the diodes must be dimensioned accordingly. All of these problems are eliminated by using the L4960 as a post regulator as shown in Fig. 7. Note that for all of the auxiliary outputs only one sec- ondary is needed, simplifying the transformer. Cross regu- lation is no longer a problem and the power dissipated in the stage depends mainly on load current and is almost unaf- fected by dropout. Moreover, the diode on the secondary can be smaller since the input cur- rent of a stepdown switching regulator is always less than the output current. Finally, short circuit protection is provided for all of the auxili- ary outputs by the chip’s internal current limiter and thermal protection circuit. Fig. 7. Using DC-DC converter circuits in place of the linear regulators simplifies the trans- former and reduces dissipation. * Giuseppe Gattavari is with SGS Microelettronica SpA How it works The SGS L4960 is a monolithic stepdown switching regulator providing output voltages from S.1V to 40V and delivering up to 2.5A output current. At the heart of the device is a regulation loop consisting of a sawtooth oscillator, error amplifier, comparator and source-sink output stage. An error signal is produced by comparing the out- put voltage with a precise 5.1V on-chip reference which is zener-zap trimmed to +2%. This error signal is then compared with the sawtooth signal to generate the fixed frequency pulse- width-modulated pulses which drive the output stage. Gain and frequency stability of the loop are adjusted by an RC network connected to pin 3. When the loop is closed directly by connecting the supply out- put to the feedback input (pin 2) an output voltage of 5.1V is pro- duced. Higher output voltages are obtained by inserting a voltage divider in this feedback path. Output overcurrents at switch-on are prevented by the soft-start function. The error amplifier output is initially clamped by the external capacitor Css and allowed to rise, linearly, as this ca- pacitor is charged by a constant-current source. Output overload protection is provided in the form of a current limiter. The load current is sensed by an internal metal resistor connected to a comparator. When the load current exceeds a preset threshold, this comparator sets a flip flop which disables the output stage and discharges the soft-start capacitor. A second comparator resets the flip flop when the voltage across the soft start capacitor has fallen to 0.4V. The output stage is thus re-enabled and the output voltage rises under control of the soft-start network. If the overload condition is still present, the limiter will trigger again when the threshold current is reached. The average short-circuit current is limited to a safe value by the dead time introduced in the soft-start network. The thermal overload circuit disables circuit operation when the junction temperature reaches about 150 °C and has hysteresis to prevent instability. ELECTRONMICROSCOPY COMES TO LIFE by Dr Jitu Shah, H.H. Wills Physics Laboratory, University of Bristol One important limitation of electron microscopes is that the conditions under which the specimen is examined make it impracticable to view living matter. Results from a technique now under development are showing rapid progress in microscopy of biological materials and are paving the way to observing the dynamics of life at high magnifications. Incidentally, it will also provide industry with a powerful tool for inspection and fault finding. A desire to see structure, forms and morphology at micro- scopical scales is inherent to the curiosity of mankind. Is was the driving force that led An- thony van Leeuwenhoek, a Dutch clockmaker, to devise a compound light microscope. Nowadays, of course, much larger magnifications can be obtained by electron micro- 11.26 elektor India november 1987 scopes. Yet optical microscopy still has a powerful advantage over electron microscopy: it can be performed on living matter without destroying it. It is interesting to note that in 1926, when the US physicist Leo Szilard suggested to the British engineer Dennis Gabor that an electron microscope might be made by assembling electron lenses, Gabor (later the inventor of the hologram and winner of a Nobel prize in physics) rejected the idea and pointed out that liv- ing specimens cannot be placed in a vacuum, which is essential for electron beam op- tics, and that energy in the focused electron beam would bum and destroy anything placed under it. In the event, the first such microscope was built by the German scientists Max Knoll and Ernst Ruska in 1931. Ad- vances since then in many branches of materials science, biology and medicine can be attributed to the use of electron microscopy. This has been duly acknowledged by the fact that Ernst Ruska shared a 1986 Scanning electron - microscope column Electron beam To vacuum pump To water reservoir To vacuum pump Tovideo amplifier Schematic diagram of the apparatus used for MEATSEM. Nobel prize in physics, so it is appropriate now to review how far electron microscopy for biological and other difficult materials has progressed. With a modern, commercially available transmission electron microscope (the type devel- oped by Ruska) we can visu- alise features and structures of only a few tens of Sngstroms. However, such an instrument requires the specimen to be thin, because an image is ob- tained by passing electrons through it. This means that sur- face features of a thick, three- dimensional object cannot be examined very well. Addition- ally, because the specimens have to be very thin, it is ex- tremely laborious to obtain three-dimensional information. These disadvantages are over- come by the scanning electron microscope, a type of instru- ment first built by the German physicist Manfred von Ardenne in 1938. Depth of focus The first commercial scanning electron microscope was made available in 1965 by Cambridge Instruments, a British firm near Cambridge. In this kind of microscope an extremely i small, focused spot of electrons is, made to fall on the surface of a specimen and scan across it in a 'raster', just as in an or- dinary television tube. The electrons interact with the specimen and release second- ary electrons from near the sur- face. (Some of the primary, incident electrons are ab- sorbed within the specimen while some are bounced back out of the surface; more about this back-scattering later.) Emit- ted secondary electrons yield information about the surface topography. In a conventional scanning electron microscope these sec- ondary electrons are collected, point by point, and used to build a picture. Deeper surface features of thick specimens can be imaged in a scanning elec- tron microscope because the depth of focus is much geater than that in an optical micro- scope, so the technique gives a much more vivid impression of three-dimensionality. It displays morphological and topological features at much higher mag- nifications than in an optical microscope. Because the in- strument is relativily easy tc use and can be combined with other analytical techniques, it has become enormously popu- lar. Magnifications available with modern instruments are only slightly less than those possible in a transmission elec- tron microscope. When it comes to viewing liv- ing matter, scanning electron microscopy has the same serious drawbacks as those pointed out by Dennis Gabor. Therefore we must be able to keep specimens in a micro- scope in a fully hydrated state, without loss of water. That is to say, they must be kept as near as possible to a living state. In any electron microscope, a well-focused beam of high- energy electrons, necessary for imaging, has to be produced and kept in a high vacuum. It cannot travel long distances in a high-pressure gaseous environ- ment without being scattered by gas atoms or molecules and losing its energy, and a badly scattered beam cannot render high resolution. This is an obstacle to electron mi- croscopy of biological material in its natural, hydrated state. For these reasons specimens are, conventionally, deliberately dried out and made stable for viewing by using procedures such as chemical fixation, dehydration and fluid replace- ment. Additionally, for scanning electron microscopy, dehy- drated specimens, which are generally poorly conducting, are coated with a thin conduct- ing layer of gold or of an alloy of gold and palladium to avoid a build-up of charge, for detailed features on charged surfaces cannot be imaged well. But these techniques cause a drastic change in interfacial tension forces, which in turn causes delicate biological structures to become distorted and even to collapse. In spite of the development of special techniques for preparing specimens it has not been poss- ible to eliminate damage com- pletely. Loss of water in a vacuum can be avoided by freezing the specimen and keeping it at a low temperature, at which , saturated water pressure is very small. In so-called cryo scan- ning electron microscopy specimens are frozen, coated with metal and transferred into the scanning microscope, where they are viewed at a low temperature. However, frozen specimens are not free from damage which takes place through anomalous expansion of water as it changes into ice crystals. Obviously the cryo technique, even if it were made free from specimen damage, could not be used to view living matter. elekior indie novembor 1987 1 1 .27 -16 " I I I I l I I l I I l I -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 BIAS FIELD STRENGTH (X 10 3 Vm 1 ) Variation of 'normalised' specimen current with field bias strength for 10-kV accelerated primary electrons. Normalisation is on the basis that the difference between the specimen current from copper and carbon at zero field strength is unity. Two regions Another approach for solving the problem of loss of water is called moist environment ambient temperature scanning electron microscopy (MEATSEM). This relies on compartmentalisation of a microscope into two regions: the first is a high- vacuum region for electron beam production and electron lense optics; the second is a high-pressure region at room temperature, to surround the specimen and prevent it from losing fluid and , gaseous constituents. (If a specimen is kept at saturated vapour pressure of water or 100 per cent humidity it will remain wet, just like clothes hanging on a washing line on a humid day.) Complete compartment- alisation can be achieved by using a window that is trans- parant to electrons but at the same time tough enough to maintain a high difference of pressure between the two com- partments. This approach has the drawback that window materials scatter the electron beam badly. To keep scattering down to feasible levels the win- dow has to be extremely thin, which means it is extremely fragile. Reliable windows with an acceptable loss of resolution are difficult to make. We have adopted an alternative method of open-window MEATSEM here at Bristol. The focused electron beam passes from the microscope to the specimen chamber via small apertures, as shown in the first diagram. Leakage of gases from the high-pressure region to the microscope column is kept to a minimum by introducing a buffer space, or intermediate chamber, between the specimen region and the elec- tron beam column. The in- termediate chamber is bounded by two walls contain- ing concentric limiting aper- tures in the planes normal to the electron beam, and it is pumped continuously so that the pressure gradients can be maintained while the micro- scope is in use. Water lost through the window is re- placed by a continuous injec- tion off water vapour in the vicinity of the specimen. To keep damage due to water loss from the specimen to a mini- mum, a sponge is introduced to the specimen chamber. The surface area of the sponge is much larger than that of the specimen, so the proportion of the water lost from the specimen to the total loss of water is very small. Scattering of the electron beam in this arrangement depends largely upon the pressure in the specimen compartment and how far the electron beam travels through the high pressure to reach the specimen. Pressure in the specimen chamber is kept at the saturated water vapour pressure, at near to room tem- perature. With careful design of the apparatus the scattering of the primary beam can be kept down to give reasonable resol- ution. We have used this open- window MEATSEM with some success: it is now possible to in- sert a fully hydrated specimen into a scanning electron microscope and keep it hydrated for a long time. Never- theless, preventing desiccation of a specimen in this way poses another problem. Additional electrons Forming an image under this conditions presents formidable difficulties. The conventional technique of constructing an image by secondary emitted electrons does not work because secondary electrons, primary electrons and back- scattered electrons ionise water or gas molecules close to the specimen and produce ad- ditional electrons. These elec- trons, which do not carry any information about the specimen surface, have a similar energy range to that of the secondary electrons emitted from the specimen, so they cannot be separated easily from the secundary electrons released from the specimen surface. Without such separation, there is a severe deterioration of the secondary emitted image. Back-scattered electrons (deflected primary electrons from beneath the specimen) also carry image information. Because they are scattered from a larger volume of specimen, the resolution achievable by their use is not as good as that obtainable by Stomata, or breathing pores, viewed by open-window MEATSEM at successively higher magnifications, which eventually -eveal a tubular structure with a slit within a stoma. 11.28 elektor india november 1987 Left-, view of part of a circuit on a semiconductor wafer showing charging effects; spread' in the darker area is due to charging. Right: A view of the same circuit after charge neutralisation. using secondary electrons. Further deterioration in the res- olution is also likely because the low-energy of back-scat- tered electrons means that they cannot be separated easily from spurious electrons. Interaction of the primary elec- tron beam with the specimen creates a charge which, in turn, generates a minute current in the specimen. The point-to- point variation of this current with scanning of the beam can be made use of for image gen- eration. With a wet specimen the current can be collected without any metal coating. A specimen-current image is also susceptible to deterioration through ambient ionisation. Our research has shown that it is possible to resharpen the im- age in a way that I shall now de- scribe. We have incorporated an ad- ditional annular electrode in the specimen chamber so that a substantial electric field can be produced at the surface of the specimen. The field has a con- siderable effect on the speci- men current, shown in the second illustration. The images are of copper grid bars on a carbon surface, and the curves represent variations of speci- men current with the strength of electric field from copper and carbon. It is thought that the electric field helps to con- duct the excess charged car- riers, produced by the inter- action of primary, back-scat- tered and secondary electrons, through the gases, which en- hances the contrast in the specimen-current image. Ap- plication of the field changes the magnitude of the specimen current rapidly at first and then more slowly as one or other plateau in the curve is reached. The image quality and contrast is re-established with the magnitude of the field applied. It is also possible to invert the image contrast by changing the direction of the imposed field. The curves shown for copper and carbon indicate that the contributions to the current by both the back-scattered elec- trons and the secondary emit- ted electrons are recovered to a great extent in the plateau regions. So, once the specimen current reaches a plateau, the contrast, sharpness and resol- I ution of the specimen-current image of an object under high pressure are substantially recovered. The image is com- parable in quality to the con- ventional, secondary emissive image of a similar object in a high vacuum. A series of pictures in the third illustration shows recovered im- ages of stomata of a fully hydrated leaf, at roughly 16 °C, at various magnifications. Stomata are breathing pores of a leaf, which automatically close and open to regulate ex- change of water vapour be- tween the leaf and the air. They are therefore suitable speci- mens to study by MEATSEM on a fully hydrated leaf. Fully hydrated internal tissue cells of animals can be imaged, too. The resolution obtained so far is limited by factors not directly related to the principle of the technique, while difficult- ies stem from the fact that the current from a typical uncoated ‘wet’ biological specimen is smaller. Results indicate that we may well see rapid progress in scan- ning electron microscopy of hydrated biological materials. In turn this will open up means of realistically assessing deterioration from other causes, such as radiation damage due to incident elec- trons and heat produced by the interaction of electrons with a living specimen. Optimistically, we may expect other advances such as low-voltage scanning electron microscopy combined with MEATSEM, which may well enable us to view live mat- ter without inevitably killing the specimen. This will not only fulfil a long standing dream of mankind but also provide a tool for observing the dynamics of life at high magnifications. MEATSEM has led to a solution of another long-standing prob- lem in scanning electron microscopy. I have already mentioned that insulating, semiconducting and poorly conducting materials are diffi- cult to image, unless coated, in a scanning electron beam in- strument because of charge build-up on the surface of the specimen. It alters the trajec- tories of primary and second- ary (emitted) electrons and the process of emission of second- ary electrons; this grossly distorts the image and is also accompanied by loss of details and resolution. Charging can also bring about electric breakdown of the material, which is particularly serious in examining semiconductor chips containing circuits and active electronic devices. With certain modifications of the open-window MEATSEM, a charge neutralisation mechan- ism can be employed to reduce or eliminate charge build-up on a surface. The final illustration shows before-and-after images of a circuit on a semiconductor wafer. The improvement was achieved by charge neutralis- ation. Potentially, the technique is a powerful tool for inspection and fault detection, and pro- mises to have other industrial uses. Cambridge Instruments, who built the first scanning electron microscope, may well be the first company to make the MEATSEM* and its associ- ated techniques available com- mercially. olektor irvdis november 1987 1 1 .29 FILTERS: THEORY & PRACTICE - 3 by A.B. Bradshaw Television, radar, data transmission, and other techniques developed during the 1940s and 1950s showed up the limitations of the image parameter theory. The higher precision and more exact characteristics required of filters from then on caused the image parameter theory to give way to the modern network theory that uses synthesis techniques and digital computers. 36 ♦ u passband jppioximation atop band approximation transition width I -*■ approximation Brlch wall (ideal) Very Hat pass band. Attenuation continues to increase in stop band. Poor pass’stop band transition . Unequally spaced ripples in pass band. Unnqually spaced ripples in Stop band. Attenuation Is defined in stop band Excite fit pass/stop band transition. Equally spaced amplitude ripples in pass band. Attenuation continues to increase in stop band. Good pass 'slop band Iransibon. E7IJMS-M The underlying principle of the network theory is to synthesize the filter from a knowledge of its voltage transfer function (voltage vs frequency charac- teristic). This is an essentially mathematical approach using the ratio of certain types of polynomials. Known as approxi- mation theory, it is a very powerful technique. Using these methods, three of the ap- proximations yield familiar transfer functions. Together with the ideal brick wall response, they are shown in Fig. 36. The development of the digital computer greatly aided in the evolution of the polynomial coefficients. This enables filter design tables to be produced from suitable computer pro- grams. The use of such tables, coupled with a slep-by-step design pro- cedure, enables the production of superb filters to the Butter- worth, Chebyshev, and elliptic function approximations. The tables refer to normalized low-pass and high-pass sec- tions. "Normalized" means that the circuit values all refer to a network impedance of 1 S at an angular frequency, a>, of 1 radian. By using standard multipliers, it is possible to translate the filter impedance to the desired value. Another set of standard multipliers enables the translation of the shape of the response to the required frequency. These two oper- ations are carried out simul- taneously and are referred to as impedance/frequency scaling. The basic structures given in the tables can be converted to high pass and a network transformation enables trans- lation into band-pass structures. These new structures are then scaled to the desired im- pedance and frequency to give practical values for the com- ponents. Once a set of filter design tables is available and the designer has familiarized himself with their main proper- ties, it is easy to produce high- quality filter designs without the need to know anything about the coefficients of real rational functions or Hurwitz polynomials. Elliptic function tables The elliptic function tables are chosen as these give the best results for a given number of components in the design (Ref. 1 ). Each table refers to a specific network. Each set of figures is given for a range of stop band attenuations (usually in 5 dB steps for the shorter network), and for a given pass-band rip- ple in dB. The figures in the columns are the actual com- ponent values referring to s is the frequency at which the specific attenuation begins; As is the specified stop-band at- tenuation; Ci — Cs are the values of the capacitors in farad; Li and La are the values of in- ductors in henry; an and wt are the angular fre- quencies at infinite attenuation. The values of the capacitors and inductors are very large because the network refers to 1 Q at 1 radian: they will be- come more practical during the impedance/frequency scaling. Note that the pass-band edge is the ripple limit and not the —3 dB point. Low-pass filter for SSB reception With reference to the last three lines in Table 3, if the pass-band edge is defined at 2.2 kHz (—1 dB), the following fre- quencies would obtain at the at- tenuations stated: SO dB: 1.407x2.2=3.0954 kHz 55 dB: 1.528x2.2 = 3.3616 kHz 60 dB: 1.674x2.2 = 2.6828 kHz If the pass-band edge were defined at 2.7 kHz (—1 dB), the frequencies at the given at- tenuations would become: id' x2«x 2,3x10’ 10’ x 2itx2.3xlO’ 50 dB: 1.407x2.7 = 3.7989 kHz 55 dB: 1.528x2.7=4.1256 kHz 60 dB: 1.674x2.7=4.5198 kHz Similarly, Ci = 181.2982 nF As regards the filter im- pedance, generally speaking, the higher this is, the larger the inductances become, and the smaller the capacitances. At relatively high impedances, the capacitors can be matched more easily by using 1% silver mica types. The large induct- ances may be avoided by elec- tronic simulation, of which more later. Using the last line of Table 3 and deciding on a filter im- pedance of 1 kQ, and a pass- band edge of 2.3 kHz, the two inductances can be calculated with inductance multiplier L‘. This multiplier is determined by £• .(design impedanceXinductance from table) 2 u (band-edge frequency) The required inductance values, Li and La are then calculated as follows: t!= l°’ x tOQ3 .=0.069405 Hx69.405 mH 2.x 2.3x10’ rU -till 1 ! 0 862 . 0.059648 H=58.6« mH 2OE > FOR ADrt=C- TO 94 i READ A£ ’get machine rode I POKE AHCOOO+ADR.VAL ~4H"-as 'write Machine coae i NEXT ADR:GCSUE 5 -80 , ............... R1JTa MACHINE CODE * i AS= INKEYS: IF CHOICS-O AND A*="" THEN 190 'select Mode i IF CHOICE -.9 AND CHOICE ■ 52 THEN 220 IF AS “1** AND AS "C AND AS - " 3 " THEN 190 i IF AS ' — THEN CHolC£=ASC(AS; IF CHOICES-? THEN LOCATE 1.6 : PRINT" - IF CH:TCE=5V THEN LOCATE l.b : PRINT" - IF CHOICE- i 1 THEN 2 .CATE 1 . 10 : PRINT"- GCSUB 090 : DEFUSR=AHCO0O DEFUSR=AHC02S COSUB 430:0070 270 A=USR C. ‘start machine cc-de F3R A -6 TO 1C ’erase arrows on screen LOCATE 1. A: PRINT " " NEXT A.GGtSUB 530 : GOTO 190 'DIRECT VOLTAGE DATA 3E.C— .06. 05.16.80. IE. CO. -F DATA 21 .AA._A .ED. El . EC .69.CE. 1A . 4F . ED. 51 .48 DATA ED.61 .C5. 12.-F.ED. 59.C9.00.00.00.00.00.0C 'SAWTu-oTH DATA 3E. 04. 06. OS. 16. 30. S£. CO. 4F. £1.01. 00. ED. 51. 48 LATA ED.c-v.CE.lA.4F.ED.5l.48.ED.6l.CB.l2.4F.ED.59 DATA I j.CB.7,..'A. 7 1 .20.21 .00. 00. ED. 51 .48.ED.6l.CB DATA :A.4F.Er.:i ,-F. ED.69.CB, 12. 4F. ED. S9.C9 define direct voltage ••••••••••• LOCATE C.16. IMFVT~ENTER VALUE. PLEASE (0-16383; ~;U POKE 4HC00A.W M7D 256 PcXE SHCOOB.W \ 256 LuCATE 0.1t FEINT SPACES 1 40 1 -.RETURN SINE -HAVE FuR U»C TC t .233 STEF .1 X* 8191. S * 8191.5 * S1H MSB= X \ 256 . LSE= X MOD 256 OUT DA.SH80 :OUT DB. 253 ‘load shittreg. a GoSVB 530 'read PIOs OUT DA.&H40 : OUT DB.K5B 'load shittreg. b GOSUB 530 ’read PlOs CUT DA.&HCO ’generate start pulse NEXT U : RETURN * read PIC status •••••••••• LOCATE 0.0 PRINT "STATUS PIO-PORTS" LOCATE 0.1: PRIN7 USING" \ V’:’’Da- " :HEXS I INP » DA ) , LOCATE 0.2. PRINT USING" V \":“Db- “.HEX*; INP(D8 ; ) RETURN . ........ jjrde select screen •••••••••• LOCATE 0.- : PRINT STRINGS t -0. : LOCATE 0.12: PRINT STRINGS <40. LOCATE 5.6 : PRINT". 1 DIRECT VcLTAGE “ LOCATE 5.8 .PRINT" _ SAWTOOTH VOLTAGE “ LOCATE 5,10: PRINT" - 2 - SINE-WAVE ~ RETURN 10 REM INITIALISE PIA 20 DA=S6822 : CA=DA» 1 : DB=CA* 1 , CB=DB* 1 30 POKE CA.0 REM SELECT ODRA 4e POKE DA, 255 : REM ALL OUTPUT 50 POKE CA.6 60 POKE CB.O : REM SELECT DR A : HEM SELECT DDRB 70 POKE DB.25S .REM ALL OUTPUT eo POKE C8.6 : REM SELECT DRB 100 REM CREATE MACHINE CODE ■* 105 FOR X=0 TO 85 110 READ A 120 POKE SC0O©*X.A 13© NEXT X 19e REM *•* MAIN PROGRAM *»* 200 CH0ICE=0: PRINT CURS ( 147 > : REM CLEAR SCREEN 205 GOSUB 1000 210 GET AS: IF AS=“"' AND CHOICER THEN 210 215 IF CHOICE^ 49 AND CHOICE 52 THEN 217 216 IF AS .”1" AND AS< -"2“ AND AS< »"3" THEN 210 217 IF AS THEN CHO!CE=A£C ; AS J 220 IF CHGICE=49 THEN GOSUB 6000 :SYS $ C03A -30 IF CHOICE=50 THEN SYS SCO0O 240 IF CH01CE=51 THEN GOSUB 3000 270 GOTO 210 3 17 IF AS • THEN CHOICE=ASC > AS ) 1000 REM SCREEN LAYOUT 1010 PRINT CHRSl 147 J : REM CLEAR SCREEN 1050 PRINT .......... 1060 PRINT " -1> DIRECT VOLTAGE" 1C7C PRINT " 2 SAWTOOTH VOLTAGE" 1080 PRINT " - 3 ■ SINE- WAVE" HOC PRINT 1110 PRINT 1120 RETURN 2000 REM SAWTOOTH 2010 DATA 120.162.0.142.00.194.169.128.141.00.222 2020 DATA 142 .2.222 . 169,64 , 141 .00.222. 173.00 . 194 2030 DATA 141.2.222.169.192.141.0.222.232.208.229 2035 DATA 254.0.194.173.0.194.201.63.208.219.169.0 2040 DATA 141.0.222.141.2.222,169.192.141.0.222.88.9b 210C REM DIRECT VOLTAGE 2110 DATA 169. 128. 141.0.222, 173. 1,194.141, 2 2120 DATA 222.169.64,141,0,222.173.0.194,141,2.222 2130 DATA 169.192.141,0,222,96 3000 REM SINE-UAVE ••• 3010 FOR U-O TO 6.283 STEP. 1 3020 X=8191 .5.B191 . 5’SINtU) 3030 MSB=INT(X/256> : LSB=X-MSB*256 3040 POKE DA.SBO 3050 POKE DB.LSB 3060 POKE DA.S40 3070 POKE DB.MSB 3080 POKE DA. SCO X 3090 NEXT U: RETURN 600C REM INPUT ROUTINE 8010 INFUT"ENTER VALUE ";B 6020 POKE SC20C. INTiB/256; 6030 POKE SC201 , B- INT ( B/256) *256 6040 GOSUB 1000 RETURN 1 1 40 Meitor india ravumber 1987 Resistors I ± 5%): Ri = 2K7 R2 = 2K5 (1K0-HK5) R3 = 82RF Rt = 620RF (470RF + 150RF) Rs = 4K7 Re = 2K2 Capacitors: Semiconductors: 1C t - 74HCT27 IC2 = 74LS132 IC3= 74HCT04 IC4 = 74HCT161 ICs = 7805 IC6 = 7905 IC7 = TDA1540' ICa = NE5534 IC9;ICio = 74HCT165 1C 1 1 =74HCT74 Miscellaneous: Xi = 10.000 MHz quartz crystal, 30 pF (ATI series res- onant: HC18/U enclosure. Ki = 20-way, double row, male header. PCB Type 87160 (available through the Readers services!. Fig. 8 Printed circuit board for building the 14 bit DAC. for the slope of the low-pass filter, and makes it possible to plj use a third-order filter with Bessel or Butterworth charac- ** ;t W teristics and yet attain more than adequate performance. Signetics-Philips • Mullard House • Torrington Place • London WC1E 7HD. For UK distributors see Infocard 509 (£F May 1987). Software Table 2 lists a program for generating a direct voltage, a sawtooth voltage, or a sine-wave with the aid of an MSX com- puter. A similar program for the Commodore C64 is shown in Table 3. It should be noted that the addresses of the input/out- put devices may have to be altered as required. In the MSX program, the loading of the D-A converter is effected in lines 470. . .510, in the C64 program in lines 3040 . . . 3080. It is regretted that software for computers other than the C64 and those in the MSX series is not available. References: m Computerscope - 2. Elektor Electronics, October 1986, p. 44. (2) MSX Extensions - 4. Elektor Electronics , January 1987. Fig. 9 Output spectrum of a DAC at a sample rate of 44.1 kHz (a) and 176.4 kHz (b) !> \o CH O O + -i i < LQ n Ql o I TnR>ln T 4+6™ o|“ 6 1 elektor india november 1987 1 1 .41 SSB ADAPTER A low-cost add-on unit that enables single-sideband reception on virtually any AM short-wave receiver. Every experienced short-wave listener knows that single side- band (SSB) transmissions can not be received unless a special detector is installed in the receiver. Unfortunately, however, an SW receiver suitable for SSB reception is generally far more expensive than an AM/FM general cover- age, radio set having adequate sensitivity and selectivity. To the dedicated SW listener, amateur and utility stations transmitting SSB signals are often more interesting than (broadcast) AM stations in view of their uniqueness, and the larger distance covered. SSB transmitters are more economi- cal than AM transmitters as regards bandwidth and power consumption, due to the ab- sence of the carrier and the second sideband. Carriers and sidebands It can be shown that the RF power contained in the carrier and one sideband of an AM modulated RF signal is redun- dant, because it is not, strictly speaking, needed to convey in- 1 1.42 elektor india november 1987 formation to the receiver. For this purpose, one sideband suf- fices. The RF output signal of an AM transmitter modulated with a single, sinusoidal frequency is shown in Fig. 1. The instan- taneous amplitude, U, of the RF carrier is a function of the am- plitude of the modulating AF tone, which can be recon- structed by drawing a line along the peak excursions of the RF voltage (the envelope waveform). Mathematically simplified, this AM signal is de- scribed by the expression UAM = Uc(fc)+/77tf c(/c + /m)+ + in Uc{fc-!rr). It is seen that the amplitude of the AM signal, £/am, is the sum of 3 terms. The first, Uc(fc), is the amplitude of the carrier with frequency A. The second and third term are of equal am- plitude, mUc, but denote sig- nals adjacent to the carrier, ie., below and above /c. The factor m represents the relative ampli- tude of the modulating signal with frequency /m. Figure 2a shows an analysis of the AM sig- nal in the frequency domain {spectrum analysis). The carrier is modulated with a single tone that gives rise to 2 side tones, each having a lower amplitude than the carrier. Modulating the AM transmitter with a composi- te AF signal, e.g. music or speech, causes two side bands rather than side tones adjacent to the carrier, and so increases the overall bandwidth occu- pied by the signal. Returning to the above formula, it is readily seen that the terms mUc(fc+f™) (upper sideband, USB) and m£A(/c+/m) (lower sideband, LSB) convey the same intelli- gence, namely the modulated signal, while the carrier, Udfc), does not convey any intelligen- ce. Evidently, the carrier and one sideband are not needed to convey information from the transmitter to the receiver, and this forms the basis of the SSB modulation method, which is sometimes— more properly— 1 Fig. 1 RF output signal of an AM transmitter modulated with a single tone. Fig. 2 Spectral analyses of AM signals. Fig. 2a: single tone modulation. Fig. 2b: modulation with a composite AF signal. Resistors (±5%|: Ri = 10K Rz;Ri = 1K0 Ra = 1 00R Rs = 4K7 Rn = 6K8 Pi = 1 00R linear potentiometer Capacitors: Ci = 200p variable capacitor Cr = 6n8 Ca:C4 = 39n Cs = 2n2 C6;Cr = 100n Ca = 1n0 Cs = 10n Cid= 22n Cu = 470n Inductors: Li= Neosid Type 7A1S* inductor assembly (see text) Lr;L3 = 270pH Semiconductors: Di;Dr- 1N4148 Ti;Tz;T3 = BF494 Miscellaneous: Si = miniature SPST switch. PP3 battery (9 VI and clip-on connector. PCB Type 87662X (not available through the Readers Servicesl. Suitable metal enclosure. * Neosid inductor assemblies are available from Neosid • Eduard House • Brownfields • Welwyn Garden City • Hertfordshire AL7 IAN. Telephone: 10707) 325011. Telex: 25423. or from Bacton Inductive Components Limited • Unit 8b • Cambridgeshire Business Park • Angel Drove • Ely • Cambridgeshire CB7 4DT. T1..T3 = BF491 51 Ik D1.D2-1N4148 LJ *t [ * u r enB 1 / ! " . > 200p 1 1: ! r L - i -1 “L ctI cel ce! ciol hie Fig. 3 Circuit diagram of the SSB adapter for AM receivers. referred to as SSBSC (single sideband, suppressed carrier). In theory, an SSB transmitter uses only, a quarter of the power of an AM transmitter for conveying the same infor- mation. In an AM transmitter, half the power is "wasted” in the carrier, the other half goes into the sidebands. The RF power of an (ideal) SSB transmit- ter drops to nought in the absence of a modulating signal. Hence an SSB transmitter has a far better power efficiency than an AM transmitter, and at the same time occupies less band- width; relatively low power SSB transmitters can, therefore, be used to cover considerable distances (maritime communi- cations, radio amateurs, etc.) without laying too heavy a claim on the available power source. From SSB to AM The operating principle of the present adapter follows from the previously discussed rela- tionship between AM and SSB. An SSB signal can be converted into AM by adding a carrier and a sideband. Both are obtained with the aid of an oscillator tun- ed to the receiver's in- termediate frequency (IF), which is usually 4SS kHz. The externally generated carrier serves as the reference fre- quency against which the (up- per or lower) sideband is demodulated. The second side- band is automatically obtained 1 Q ©-^ 1 Q 7BB2X\ T tl psr/. y | i ohh> in this process, so that a double sideband AM signal is available for demodulating. The combi- nation of the AM receiver and the SSB adapter is, understan- dably, not up to a real SSB com- patible receiver with its special, narrow band, IF section. None the less, the results obtained with the present add-on unit are satisfactory for relatively strong, interference-free, signals. Circuit description The SSB adapter is a simple cir- cuit, shown in Fig. 3. Transistor Ti oscillates at 4SS kHz with the aid of parallel tuned circuit Ci-Li. The oscillator signal is raised and filtered in a cascode amplifier set up around T 2 and ^6 6 0 O 1 Fig. 4 Printed circuit board for the SSB adapter. elektor india november 1987 1 1.43 Fig. 5 The Type 7A1S inductor assembly from Neosid. 1: screening can. 2: ferrite cup. 3: iron dust core. 4: ABS former and base. Ti The amplitude of the output signal is made variable with Pi, enabling optimum perform- ance with any receiver. The adapter’s output signal is con- nected direct to a length of in- sulated wire, wound as 1 or 2 turns around the receiver (in- ductive coupling). The adapter is fed from a 9 V battery. Construction and alignment The printed circuit board for the SSB adapter is shown in Fig. 1 4. Construction is straightfor- ward with the possible excep- tion of inductor assembly Li — see Fig. 5. Viewed from under- neath, the base of the ABS former in the Type 7A1S as- sembly has 5 pins, 3 at one side and 2 at the other. Inductor Li is connected to the latter 2 pins. Close-wind S3 turns of 00.2 mm (36 SWG) enamelled copper wire onto the 2 sections of the former, and make sure that the ferrite cup (part 2 in Fig. 5) can be fitted on top. Secure the winding with a piece of Sellotape. Check the continuity at the base, and fit the former onto the PCB. Carefully slide the screening can over the former, then push-fit and solder its mounting tabs in the holes provided. Make sure that the top end of the former fits snugly in the hole in the top of the screening can. Tuning capaci- tor Ci and level control Pi are fitted as external components. It is recommended to fit the SSB adapter in a metal enclosure to prevent spurious radiation. Set variable capacitor Ci to the centre position, and Pi to maxi- mum. Connect the coupling loop around the receiver to the adapter output. Tune the re- ceiver to an AM broadcast station, and switch the adapter on. Adjust the core in Li with a non-magnetic trim tool until a whistle (beat note) is heard in the receiver. Lower the fre- quency of the beat note by ad- justing Li, until it is no longer audible (zero beat tuning). Switch off the adapter, and tune the receiver to an SSB station. Switch the adapter on again, and adjust Ci and Pi until the speech becomes intelligible. B DRILL SPEED CONTROL LI Most drill speed controllers suffer from one or more drawbacks. These include poor speed stability, excessive instability at low speeds, and high power dissipation in the series resistor used to sense motor current. The circuit described here suffers from none of these drawbacks, and in addition is extremely simple. The mains input is rectified by D1 and dropped by Rl. The current drawn by Tl can be controlled by means of PI, thus also controlling the DC voltage that appears across C2, and hence at the base of T2. T2 is connected as an emitter follower, and the voltage appearing at the cathode of D3 is about 1.5 V less than the base voltage of T2. Assuming that the motor is turning but that the triac is turned off, the back e.m.f. generated by the motor will appear at the Tl pin of the triac. So long as this voltage exceeds the cathode voltage of D3 the triac will remain turned off, but as the motor slows down this voltage will fall and the triac will trigger. If the load on the motor increases, thus tending to slow it down, the back e.m.f. will fall more quickly and the triac will trigger sooner, thus bringing the motor back up to speed. Since the triac can be triggered only on positive half-cycles of the mains waveform the controller will not vary the motor speed continuously from zero to full speed, and for normal full-speed running SI is included, which turns the triac on permanently. However, the circuit exhibits good speed control characteristics over the important low speed range. LI and Cl provide suppression of r.f. interference generated by the triac. LI can be a commercially available r.f. suppression choke of a few microhenries inductance. The current rating of LI should be from two to four amps, depending on the current rating of the drill motor. Almost any 600 V 6 A triac can be used in the circuit. 11.44 elektor india november 1987 SWITCH-MODE POWER SUPPLIES Recent advances in power semiconductor technology and inductive components have boosted the use of compact, high efficiency power supplies of the switch-mode type. Now SMPSs of various power ratings are becoming widely available at reasonable prices, it seems timely to focus on their design principles and practical aspects. Most electronic circuits can not work without a power supply of some kind. The basic mains supply consists of a trans- former, a rectifier, a filter (smoothing/reservoir capaci- tor), and a linear control circuit (regulator) for adjusting the out- put voltage to the desired value. It may be argued that the basic power supply has a number of important disadvantages. For relatively high powers, the mains transformer is often bulky and expensive, and the same goes for the smoothing capacitors). Moreover, the product of the voltage drop across the regulator and the current consumption of the load forms dissipated, and therefore wasted, power, which results in a very low overall ef- Rg. 1 The 3 basic configur- ations of a switch -mode power supply. ficiency, especially at relatively low output voltages. Not sur- prisingly, in the rapidly ex- panding world of microelec- tronics there arose a growing need for a high-efficiency power supply. This need was met by the switch-mode power supply (SMPS), in which the output power is not regulated continu- ously, but pulsed at a relatively high frequency. An output filter is included for smoothing the supply voltage. The filter components can be kept small thanks to the high frequency, and the same goes for the (toroidal) transformer if galvanic insulation is required. Basic configurations A switch-mode power supply is essentially a DC-DC converter. The 3 basic circuit configur- ations are shown in Fig. 1. The flyback circuit works as follows. A magnetic field builds up in the inductor as long as the switch remains closed. When the switch is opened, the induc- tor functions as an energy source. The voltage across the inductor is reversed, and the conducting diode passes the energy to the reservoir capaci- tor. Note that the output voltage is reversed with respect to the input voltage. The forward converter does not reverse the polarity of the input voltage. The capacitor is charged via the inductor when the switch is closed. The differ- ence between the input and the output voltage is available on the inductor. In contrast to that in the flyback converter, the switch is closed when the ca- pacitor is being charged. When the switch is opened, the mag- netic field of the inductor is weakened via the flyback diode. The switch is, of course, i a power transistor, and the di- ode affords protection against the induced voltage. In a for- ward converter, the input volt- age is higher than the output voltage. The third basic configuration is referred to as boost or step up converter. This circuit in- creases the input voltage, and is functionally similar to the flyback converter. Energy is stored in the inductor when the switch is closed. When the switch is opened, this energy is supplied to the load at the out- put via the diode. The continuous and discontinuous mode Two modes of operation can be distinguished, depending on the current in the inductor— see Fig. 2. After closing the switch, the current in the inductor in- creases linearly up to a specific maximum value (Ui = constant). After opening the switch, the current decreases linearly. The circuit operates in the discon- tinuous current mode if the cur- rent is nought in every period. The capacitor supplies the load current during the remainder of the period. The discontinuous mode is characterized by the good response of the closed regulation circuit to fluctuations in the input voltage (line regu- lation), and the output load (load regulation). There is no energy in the inductor at the start of each period, and regu- lation can, therefore, take place on a period-to-period basis. It can, in fact, be argued that the inductor is not present in the regulator circuit. The maximum phase shift of 90° in the buffer capacitor ensures the stability of the closed regulation circuit. A disadvantage of the discon- tinuous mode is the relatively high peak current carried by the power switch. Flyback and step up converters usually operate in the discontinuous mode. In the continuous current Fig. 2 The discontinuous (a) and the continous current mode (b) dif- fer in respect of the current carried by the inductor. elflktor india november 1987 1 1 .45 3 Fig. 3 A traditional soft iron core (background) and a modern ETD ferrite core of equal power rating. mode, the current through the inductor does not drop to nought at the end of every period. The ripple current in the inductor is small relative to the load current, and this re- quires a fairly high self- inductance. The buffer capaci- tor, on the other hand, can be kept relatively small. The favourable shape factor of the current through the power tran- sistor and the diode makes the continuous mode eminently suitable for high power appli- cations. The response to load fluctuations is, however, worse than that of a circuit in the discontinuous mode. Each change in the output load cur- rent requires a corresponding change in the direct current through the (relatively large) self-inductance, and this pro- cess may take several periods to complete. It is not possible for a system to automatically switch from con- tinuous to discontinuous oper- ation, or vice versa, because this would cause a con- siderable change in the open loop transfer characteristics, giving rise to instability of the closed regulation system. This means that the load current of a system in the continuous mode should be higher than half the peak-to-peak value of the ripple current in the inductor. Forward converters usually operate in the continuous current mode. Off-line operation In many cases, the input voltage is a rectified and smoothed 11.46 elektor india november 1987 voltage obtained direct from the mains. The direct voltage so obtained (approx. 33S V at a 240 VAC mains supply) is rarely used for converting down to, say, 12 V, in view of the resultant low duty factor, and the need for large self-inductances. Also, a direct connection to the mains is dangerous, and nor- mally not permitted. This calls for a (ferrite) transformer, which, in an SMPS, has the ad- vantage of being much smaller than a soft iron type used in the traditional 50 Hz mains supply (see Fig. 3). There are, however, a number of important con- siderations as to keeping the losses of the core material within acceptable limits. Ferrite is used instead of laminated iron, and offers a number of ad- vantages. The construction of a ferrite core is relatively simple, and ferrite is a light and in- sulating material. The turns ratio of the ferrite transformer enables converting the high input voltage down to a value close to the desired out- put voltage. The conversion in- creases the duty factor, and hence reduces peak currents in the power transistors. Transformer circuits The simplest configuration of an SMPS is the single transistor flyback converter shown in Fig. 4a. This circuit is well-known in low power supplies with an out- put rating up to about 250 W. In this application, the transformer is more properly referred to as a coupled self-inductance, because it assumes the function of the inductor shown in Fig. la. The forward converter in con- tinuous mode is more suitable for feeding relatively heavy loads. The most commonly found version is based on a single transistor and a demagnetization winding in the primary circuit— see Fig. 4b. The transistor must be able to handle twice the input voltage. The demagnetization winding can be omitted if the circuit is extended with a transistor and a flyback diode as shown in Fig. 4c. In this circuit, the transistors need only withstand half the voltage. They are, however, driven with respect to different potentials, just as in the bridge circuits to be discussed. Half or full bridge circuits are used mainly for high power ap- plications. The full bridge variant is suitable for very heavy loads thanks to the fact that the effective input voltage is doubled. The last variant, shown in Fig. 4g, is also a bridge circuit, based on a centre-tapped transformer that enables the transistors to be driven with respect to a com- mon reference potential. Bipolar transistors as well as power FETs can be used in the primary circuit. Bipolar tech- nology is suitable for switching frequencies up to 50 or 100 kHz. Power FETs are faster, and can be used at higher frequencies without running into excessive switch losses. Currently, the maximum usable frequency is about 1 MHz, and power FETs are expected to become predominant in SMPSs in view of the ever increasing switching frequencies. Power FETs for relatively high voltages are, however, still quite expensive, and more attractive for use in countries with a 117 V mains supply, such as the USA. Core saturation Any transformer winding forms a self-inductance, and the average voltage across it should, therefore, be nought. When this is not so, the remain- ing direct voltage causes an lin- early increasing direct current until the core is saturated. The Fig. 4 Various configurations of the switching power stage in an SMPS. Fig. 5 Functional diagram of a SMPS with the control electronics located at the secondary side la) or the primary side (b). H-fi61d, and with it the current, then increases exponentially, in accordance with Faraday's law of constant increase of the mag- netic flux per unit of time, This effect must be prevented because it can lead to destruc- tion of the primary circuit. In the circuits of Fig. 4b and 4c, the field in the transformer core is weakened with the aid of the flyback diode(s), but only as long as the duty factor remains below 50%. Problems owing to permanent magnetization are not expected to arise in the cir- cuit of Fig. 4d, where the coup- ling capacitor ensures the absence of direct current through the primary winding. The situation is more complex in Fig. 4e.Although the primary winding is AC coupled, a direct voltage may still exist at the junction of the capacitors. This positive or negative potential may arise from less than perfect (i.e., unbalanced) driving of the power transistors, which may have different recovery times also. Unbalancing of the primary circuit can be prevented with the aid of a com- pensation winding and 2 fly- back diodes. A coupling capacitor for blocking the magnetizing i direct current is rarely used in the circuit of Fig. 4f, because this is rated for very high power. Both the positive and the negative current in the primary winding are measured, and any difference between them is compensated by con- trolling the duty factor. This safety measure can be applied to the circuit of Fig. 4g also. Voltage control In a switch-mode power supply, the output voltage is measured, compared to a reference, and kept constant by controlling the duty factor of the drive signal applied to the power switches (i.e., transistors). The regulating effect of the control circuit depends on the open-loop characteristics of the system. The simplicity of the flyback converter makes it less suitable for many purposes, since the duty factor depends primarily on the load at a constant output voltage. Good voltage regu- lation requires a high amplifi- cation of the measuring and control circuit. In a forward converter, the voltage control circuit need not have a strongly regulating effect because in i essence the output voltage depends only on the turns ratio of the transformer (assuming a constant input voltage). There are 3 basic types of voltage control system: • Direct duty factor control. The error signal is amplified, and drives a pulsewidth modu- lator, which in turn adjusts the duty factor as required. A high overall amplification is needed, at the cost of some stability — notably in the case of con- verters operating in the con- tinuous mode. • Voltage feedforward. This is the most commonly used system. Preregulation of the duty factor is implemented as a function of the input voltage, enabling the output voltage of the open loop system to be made independent of the input voltage. The control circuit is, therefore, only required for compensating load fluctuations. The preregulation system im- proves the line regulation, and so ensures sufficient suppres- sion of hum. • Current mode control. A sec- ond control circuit (inner loop) inside the voltage control circuit (outer loop) enables switching off the power transis- tor at a more or less fixed peak value of the current. The effect so obtained is the (quasi) disap- pearance of the inductor from the output filter. The whole system is then essentially a first order network with the capaci- tor in the output filter as the only phase changing element. The stability of the whole supply, as well as the response of the closed system to fluctu- ations in the input voltage and the load current, is excellent. High power forward converters of the continuous type are often equipped with a current mode control for obvious reasons. Location of the control circuit The voltage control circuit can be located at the primary or the secondary side of the supply. A circuit at the primary side (Fig. 5b) makes it possible to drive the power stage direct from the IC, while it is a relatively simple matter to implement circuits for primary current monitoring and preregulation. A disadvantage of the primary location is the need for an insulating device in the control loop for transmitting an analogue signal from the primary to the secondary side. This is usually done with the aid of an optocoupler. Only the er- ror signal is transmitted to rule out instability of the output voltage owing to ageing effects in the optocoupler. A control circuit at the second- ary side (Fig. 5a) enables direct coupling of the voltage control circuit. Also, it allows the use of the the reference circuit built into many of the currently available integrated SMPS con- trollers. The FWM signal is fed to the power stage via a fast op- tocoupler, or a special pulse transformer. Differences in the drive applied to several power transistors are relatively simple to monitor and correct, but the primary location makes it diffi- cult to keep tabs on the primary current. n c sync »M U 0 («ttlUl*d| control amplifier output feedback input slow start current limit output & max. >S inhibit Fig. 6 General block diagram of an integrated SMPS controller. elektor india november 1987 11.47 Whatever control circuit is used, it must have its own (start) supply. The self-generated voltage is, of course, only usable after the supply is fully operational. SMPS controllers A wide variety of integrated cir- cuits is currently available for controlling switch-mode power supplies. These ICs are essen- tially very similar, and a general block diagram is therefore given in Fig. 6 to explain their operation. The pulsewidth modulator is composed of a sawtooth gener- ator, a voltage comparator and a set-reset bistable. A second in- put -on the comparator is con- nected to the output of an opamp that amplifies the differ- ence between the real and the required (set) output voltage. An accurate, and temperature compensated, voltage refer- ence is often included for ad- justing a specific output voltage. The remainder of the circuits in the chip have auxiliary func- tions, and serve for various types of protection. Voltage feedforward or current mode control is possible by changing the slope of the sawtooth signal. Special control inputs make it possible to set a duty factor of nought. An analogue input is activated above a predefined voltage level, and can be used for making a 2-level short- circuit protection. When the output current reaches the first level, the duty factor is held constant, and the circuit sup- plies a constant, maximum, cur- rent. The duty factor is made nought above level 2. A digital input enables remote control of the supply (computer- controlled test sites, etc.). An input for setting the maxi- mum duty factor is standard on most SMPS controllers. Prop- erly driven, it prevents satu- ration of the transformer core, and hence an exponentially ris- ing primary current. This safety measure is especially useful for supplies operating in the con- tinuous current mode. In these, the duty factor has a tendency to rise to the maximum value at each change in the output cur- rent, in an effort to correct the direct current through the in- ductor in the output filter in ac- cordance with the new load. 11.48 elektor India novembe* 1 987 Fig. 7 Circuit diagram of a 5 V; 20 A; 50 kHz supply (courtesy Siemens). The multi-voltage supply Additional supply voltages are fairly simple to implement in a SMPS with the aid of appro- priate auxiliary windings on the secondary of the transformer. In computer equipment, the usual combination is a powerful 5 V section, and auxiliary +12 V supplies. Voltage control is usually only effected on the + 5 V rail. If the magnetic coup- ling between the secondary windings on the transformer is sufficiently tight— this is typical of a well-constructed trans- former— the output voltage of the auxiliary windings is regulated along with the main supply, at acceptable accuracy. Losses Switch-mode power supplies are known mainly for their high efficiency. In spite of this, some power is, of course, wasted. • Switching losses in the power transistors. Faster switching— e.g. with the aid of a speed-up capacitor— keeps these losses acceptable. The power loss incurred in the con- ductive transistor is relatively small, especially at high input voltages. • Transformer losses can be classified as copper or core losses. At switching fre- quencies below 100 kHz, cop- per losses are the main consideration in finding the op- timum specifications of the transformer in the SMPS. Also, t care should be taken to counter a considerable skin effect, ; which reduces the effective diameter of the copper wire as the frequency increases. Many SMPS manufacturers use litze wire for their transformers to prevent losses arising from the skin effect Core losses are due to eddy | currents and hysteresis of the ferrite material. They depend on the so-called Sux density sweep, and the frequency. Manufacturers of ferrite cores can supply graphs to establish the maximum permissible core losses as a function of the ther- mal resistance of the core, and other parameters. Core losses form the crux in designing a transformer for use in an SMPS, • Rectification and filter losses. These become more serious at relatively low output voltages (5 V), and are mainly due to the forward voltage drop across the diodes. Schottky diodes are often used in view of their low forward voltage drop and good switching characteristics. Some power is also wasted in the in- ductor as part of the output filter. A practical circuit The circuit diagram of a typical switch-mode power supply is shown in Fig. 7 (Siemens Appli- cation). Buffer capacitor Ci is fitted at the output of a bridge rectifier, which is fed from a mains filter. Power resistor Rv limits the peak charge current when the supply is switched on. The type of transformer used makes clear that the circuit is a for- ward converter. The primary and secondary winding (Nz; Ns) are in phase, as indicated by the dots. Auxiliary windings Ni and N3 serve to demagnetize the core. The dotted line in the core indicates the use of an elec- tromagnetic shield. The con- figuration of the secondary output filter shows that the supply is designed for oper- ation in the discontinuous cur- rent mode. A relatively small self-inductance is used in con- junction with a large buffer capacitance to ensure sufficient output power when the induc- tor carries no current. The out- put voltage is divided and compared to a 3 V reference. The error signal from the oper- ational amplifier is fed to the TDA4718-based primary control circuit via an optocoupler. Components Ct and Rt define the switching frequency of 50 kHz. Provision has been made to set the maximum out- put current (pin 9). The maxi- mum primary current is monitored with the aid of the network connected to pin 8. The capacitor fitted at pin 15 of the controller ensures a gradual increase of the duty factor after power-on (soft start). The input voltage is checked via pin 6 and 7. The duty factor is made nought when the input voltage is either too low or too high. Voltage feedforward is im- plemented with the aid of Rr, connected to pin 2. The switch 1 Fig. 8 A compact switch-mode power supply. signal at the output of the con- troller is fed to the power MOSFET via a number of paral- lel CMOS buffers contained in a CD4049 package. The control circuit is fed from the mains via a capacitive voltage divider. Figure 8 shows a compact SMPS fitted on a printed circuit board. The arrows point to the follow- ing, essential, parts: (1) mains filter; (2) primary rectifier; (3) buffer capacitor for primary voltage; (4) switching transistor; (5) pulse transformer for base drive; (6) SMPS controller for pulsewidth modulation; (7) (sec- ondary) voltage reference and error amplifier; (8) ferrite core transformer; (9) primary flyback diode for weakening the trans- former field; (10) inductor in output filter; (11) secondary rec- tifier and flyback diode; (12) output capacitor. Further developments The scope of this introductory article does not allow a detailed discussion of all the technical considerations that go into designing a switch-mode power supply. In a forthcoming issue of Elektor Electronics the subject will be reverted to in the context of a construction project. The theoretical aspects of the SMPS have been known for some time, but it was not until the coming of fast power tran- sistors, integrated controllers and new ferrite materials that serious development of the SMPS was launched. Ever higher switching frequencies make it possible to reduce the size of the secondary filter, but at the same time pose a real dif- ficulty as regards elec- tromagnetic interference (EMI) due to the often large number of strong harmonics and other spurious products. It is with this in mind that there is a growing interest in free-running sup- plies, in which the currents are sinusoidal rather than rec- tangular. Meanwhile, supplies of the types discussed in this article are constantly reworked and enhanced to make the cur- rent consumption from the mains sinusoidal. This ensures less interference on the mains, a higher efficiency of the recti- fier thanks to the more favourable shape factor of the current, and reduced peak cur- rents in the buffer capacitor(s). TW For further reading: • High frequency power transformer and choke design. Part 1...4 incl.\ Philips Technical publi- cation, September 1982. • Electronic Components & Applications. Vol. 2; No. 7; Philips, November 1979. • Unitrode Switching Regulated Power Supply Design Seminar Manual', Unitrode, ed. 1986. • Schahnetzteile (SNT), Technik und Bauelemente; Siemens Technische Mit- teilung No. B9-B3269. • Steuer und Uber- wachungsschaltungen fuer moderne Schaltnetzteile (SNT), TDA47xx Familie ; Siemens Technische Beschreibung No. B/3132. • Integrierte Schaltnetzteil- Steuerschaltungen, Funktion und Anwendung; Siemens Technische Beschreibung No. B1-B3U6. • SGS power supply appli- cation manual', July 1985. • Various databooks and ap- plication notes: Intersil; Mullard; SGS; Siemens; Thom- son; Unitrode. eleklor India november 1987 1 1 .49 THE POSITIVE IMPEDANCE CONVERTER by A.B. Bradshaw One of the very practical means of simulating inductance in electrical circuits is by the use of gyrators. The positive im- pedance converter is a member of this family. Its main use is to replace wound inductors in AF circuits, particularly where these have large values, or to simulate coils at very low frequencies. The positive impedance con- verter makes use of operational amplifiers: two are needed to simulate a grounded inductor; four are required to simulate a balanced inductor. But only four opamps are needed to simulate a node, containing balanced and grounded induc- tors, as will be shown later. Where simulation of grounded inductors is used, the opamps should be operated from balanced power supplies. The writer has used the PIC as a circuit element for a number of years in the design of high- performance AF filters. The Q values obtained with these devices is very much higher than that of the wound equiv- alent. Instability problems are rare. Although the frequency response of the opamps usually limits the operating range of PICs to about 40 kHz, this is ad- equate for most AF and even a number of data filtering re- quirements. The circuit diagram of a typical PIC is shown in Fig. 1. The input impedance of this circuit ap- pears as a pure inductance when the output is grounded through Ro. Resistors Rt, Ri, and Ri, as well as capacitor C, are normally 1% types. Resistor Ro is used as the inductance setting component. The general symbol of a PIC is shown in Fig. 2, but in practical circuits, when used as a circuit element, it is usually indicated as in Fig. 3. Analysing the PIC The analysis of gyrators is nor- mally performed with the aid of matrix algebra and the formal nodal analysis of network theory. The formal approach has a lot to offer as regards generality, but it sometimes, tends to obscure the practical operation of the circuit. Because of this, the writer has adopted an approach which will be familiar to most readers. Assuming that opamps are perfect, the analysis can be simplified considerably. In an ideal opamp, ■ the voltage gain is infinite: Avo— ■ the input resistance is in- finite: ■ the output resistance is zero: . r ou; = 0 ; ■ the bandwidth is infinite: BW = oo; ■ there is zero input offset voltage: £o=0 if Em-0. Since the voltage gain is infinite, any output voltage is the result of an infinitely small input voltage. In effect, therefore, the differential input voltage is zero. The preceding assumptions are used as axioms in the following. For the purposes of examining the operation of a PIC, it is redrawn in Fig. 4. Some of the voltages and currents are shown twice to emphasize the circuit action. (V,-Vi)/R 3 =h Eq. 1 (Vi-V* )/J?,=A Eq. 3 Since the input resistance of the opamps is infinite, /o=0, whence h=h and h-Im From Eq. 3: V,-V>=R'h=RJm Dividing both sides by Ri gives (Vi-VV )/Rz=ImR,/R 2 Eq. 4 Since h=h, Eq. 1 is equal to Eq. 2, so that (Vl-V 2 )/i?3 = (Vl-V4)/J? 2 = /inf?i/i? 2 Eq. 5 To remove Vz from Eq. 5, con- sider the middle section of Fig. 4, which for convenience’s sake is reproduced in Fig. 5. (V,-V*)IRz=h Eq. 2 4 Ri Also, since lo- 0, UIVilRo Combining these expressions gives V,-Vz= Vz/jaCRo Eg. 6 Dividing both sides of Eq. 6 by R] gives (Vi — Vz)/ Rz— Vi/jwCRoRz in which the left-hand term is also that of Eq. S, whence LrR : ! R?- VWjwCRoRi from which Vi has been removed. Cross multiplying this last ex- pression yields V\/Iin~j(i)CRc*RlRz/Rl Since Vi = Vm, Vinllin—Zm = jtx)(CRlRl / Ri}Ro which is the expression for a pure inductance in which L=(CRMRi)Ra UK=CRzSilRi, L-KRo K is called the conversion factor of the PIC This completes the basic analysis of the PIC without the need of anything more than elementary AC theory and algebraic manipulation. Practical applications The numerical values of J?i, Ri, and Rz affect the signal hand- ling capabilities of practical opamps and are, therefore, a compromise. Typical values are: A?i = 270R 1% J? 2 = 5 k 6 1 % J?a = 10 k 1 % C = 10 nF 1% silver mica (two 5000 pF types in parallel). These values enable the com- putation of K: K=CRiRz/Ri = = 2 70 x 10 4 x 10 x 10‘ s /5.6 x 10 3 = = 4.8214x10-® In practice, the author used an MC1458 operating from a ± 12 V power supply, which has a 7T=5.7155xl0' 6 . The departure from the ideal K value is due to the approximations used with the opamps: this is not a real problem in practice. The com- plete circuit of the PIC with the values stated is given in Fig. 6. Checking the value of K in practice Build the circuit of Fig. 6 and bring it to series resonance with the aid of a test set-up as 8 2 C 2 C o— 1| II — O c II -ST— II O _ G>- PIC 1—0 O- -o o- -o -o — o o- 0 ’“H PIC, cJ. <^> ¥ -o o o -o Li is simulated by PICi + PIC 2 (R