CONTENTS May 1990 Volume 16 Number 178 {In next month's issue: + Mini EPROM viewer + Electronic load * Automatic power down + Four-sensor sunshine meter + Adaptation of Alar fax {or IBM PCs + Remotely controlled stroboscope MIDI master keyboard + PLL sinewave generator Electronic fuses Front cover Data relay satelite play a Vial ole inthe space con: ‘munications network, but thy are so huge and con: plex that hey are in danger of becoming unmanageable ‘The tiny detector unt shown in the photogragh, designed by SIRA, a leading British research and development compen, wil play a vital role ina projet aimed at reducing both size and complexty Her, itis being aligned and calibrated ater manuiacture ‘The detector, developed in colaboration with British ‘Aerospace is pat ofa track: ing sensor for a project nown as sux, due to be Faunched by the European ‘Space Agenoy in the early 4990s, The objct of this isto develop optical intersatelite communications links (st) that wiln future be used to transfer data and commands between the data relay satel lites and operational satellites, inlow orbit sts wil also be used 1 lnk pairs or groups of gec-staionary communica- tions satelite. Optical communications are more efoent than a conver Uional microwave beam, which means that satalites wil nt need tobe 80 large and will be able to cary far smaller antennas than ai pre- sent SIA LTD, South Hil, Cisl hurst BR7 SEH, England! telephone 081 467 2636, 1B 20 60, 30 ‘LEADER Copyright and the SAVE decoder AUDIO & HI-FI PROJECT: Hom loudspeaker by T. Giffard PROJECT. Intro-scan for CD players by J. Ruffell COMPUTERS & MICROPROCESSORS: PROJECT: PC servicing card an ELY design PROJECT: Centronics A-D/D-A converter by J. Ruttell INTERMEDIATE PROJECT Acoustic temperature monitor by J. Rutfell RADIO & TELEVISION The multi by J. Buiting PROJECT: An experimental all-waveband ferrite rod anterma by Richard Q. Marris. G2B2Q SCIENCE & TECHNOLOGY. Image segmentation by MS. Kishore MAC concept ‘TEST & MEASUREMENT PROJECT: Budget sweep/tunction generator by T. Wigmore PROJECT: Transistor characteristic plotting by S. Aaltonen ‘MISCELLANEOUS INFORMATION Electronics scene 14; Events 36; New books 36; Letters 63: Switchboard 63; Readers services 65; Terms 66: Buyers guide 74; Semi-display adverts 74; Index of advertisers 74 Profile of Maplin Electronics p. 55 ‘A-DID-A converter - p. 41 Sweepitunction generator -p. 50 BLEKTOR ELECTRONICS MAY 1990Telex 917000 LPC G) “Fax. 081-874 9153 (Nationa) F286 18740153 Iiematonad ‘Advertising: PRE Lined “2 Woleeey Torre ‘CHELTENMAM GL50 1TH Telepona (0242) 510760 | Fax (0242) 228626 -Europeon Orfces: | Posts 75 5100 AB BEEK |The Neteriands Theprone: x31 4490 99444 “oles: 56517 (lati), Fax 31 4400 7016) Managing Director. MJ. Lancmnan Overseas eaions: FEDERAL GERMANY ELEKTOR ELECTRONICS MAY 1999) COPYRIGHT AND THE SAVE DECODER In our leader of February 1988 we warned of the possible consequences of some dubious proposals in the Copyright, Designs and Patents Bill then going through the House ot Lords. The Bill became the Copyright, De signs and Patents Act [988 and came into force ow 1 August 1989. We felt (and still feel) that jn panicular the protection of « satelite transmission provided from a place outside the United Kingdom (now Section 299) should be removed. Fortu- nately, this section requires the making of an Order in Council Furthermore, we believed (and sll do} that Section 298. giving a person who makes charges for the reception of certain satellite broadcasts the same rights und remedies as a copyright owner has against a person who publishes any information whieh is ealeulated 1o enable oF assist persons to receive these broadeasts without payment, should be amen: ded to exclude researchers and experimental: ists who receive the broadcasts in the further: ance of science and technology, After all radio and TV would not be where they are today without the pioneering work of thou- sands of radio and TV amateurs and experi mentalists ever since Marconi first spanned the Atlantic Ocean by wireless. Early in 1989 we had purchased the copyright of an experimental decoder of orig. inal design for the BBC-TV Europe transmis sions. It should be noted that this design was confirmed dissimilar to the official decoder built and distributed for BBC Enterprises by ‘Space Communications Sat-Tel The article was not ready for publication until early August 1989. Since by then the taken effect. we new copyright ct ha sought advice and were recommended not 10 publish the article until the High Court in London bad given judgment in a case in whieh BBC junction prohibiting the sale of unauthorized decoders capable of descrambling the BBC TV Europe signal transmitied from the Inte sat V satellite, This was to be the first impor ant ease t0 lest the new Copyright, Designs sand Patents Act 1988, “The case was heard last November and its ‘outcome threw the satellite TV industry into disarray; Mr Justiv Seott ruled that “there is rho copyright in waves in the ether” and lifted 1an injunction on the sale of pirate decoders. ‘The ‘pirate’ in this case was HieTech Xtravi- sion, which had been selling decoders in Spain, The case was widely reported in the general and technical press, both in the UK and abroad. In view of this judge! sign was (and is) not intended for commercial gain, but aimed af the radio/TV experimental ist and students and technicians as an ad to mecprises Was seeking an in- understanding modern technology. We con: cluded that publication of the article was legit Imaie. Part | of the article was contained in ‘our February 1990 issue, which went t0 press in early December 1989, and was on sale by mid January 1990, By the time the March issue. containing Part 2, had gone to press in January. some thing had happened which had not been given such wide publicity and of which we were consequently and unfortunately not aware, A few days before Christmas BBC Enterprises had been sucessful in its appeal against Mr Justice Scott's ruling. The Court of Appeal de- {ded that the making, selling or importing of decoders which had not been authorized by the broadcaster or senider of enterypted trans: missions was contrary 10 Section 298 of the Act. Tt was only when our March issue was about to be distributed in February that we were advised by solicitors acting for BBC E: terprises nol to continue publishing material intended for the construction of experimental decoders of rut ihe risk of prosecution under Section 298 (2) (b) which says that anyone who “publishes any information that is calcu lated fo enable or assist persons t0 receive programmes or other transmissions when they are not entitled to do so” commits an infringe: ment of copyright 1, should be noted that the protection af- forded by Section 298 is not dependent on the person providing the service making a charge. Section 29X (1)(b) refers to a person who ‘sends encrypted transmissions of any other description” and it way partly for this reason that the Court of Appeal construed “when they ‘are not entitled to do so” as meaning Without the authorization of the person who either makes. charges or senids encrypted transmis- Since we firmly believe that the law should be upheld, even though. in our opinion, it eon: tains flaws, we withdrew the March issue and reprinted it minus Part 2 of the decoder article This as inevitably meant a delay in the availability of the March issue, which finally ‘went an sale almost three weeks late. It also ‘meant that the April issue had to be delayed so as 10 give the March issue a reasonable expo- sure in bookstalls and newsagents. We hope that in due course the act will be amended by other court cases so as to restore to radio/TV experimentalists the right to de- sign and build any type of equipment in the furtherance of science and technology. In the ‘mean time, however, we must warn readers Who have the wherewithal 10 construct the de- coder from the information in our February 1990 issue that, if they do so, they will be in breach of Section 298 (2) (a) of the Copyright Designs and Patents Act 1988, .CONSUMER ELECTRONICS AND THE EEC's ANTI-DUMPING POLICY Before the EEC imposes anti-dumping du- ties on imports from outside the Commu- nity—pushing up prices to business and domestic consumers—the costs to the con- sumer should be calculated, and consumer organizations should be consulted. This is just one of a series of recom- mendations in a paper published by the National Consumer Council: Consumer Electronics and the EEC's anti-dumping policy, ‘Written by Nigel Grimwade, senior lec- turer in economics in the Department of Business and Finance at South Bank Poly- technic, the paper Counts the cost to con ssumers and business of anti-dumping mea. sures taken by the European Commission to protect European electronics manufac- turers from competition from the Far East. Itis the first in a series commissioned by the National Consumer Council as part ‘ofa study of how intemational trade poli- cies affect consumers, and was published by the NCC to stimulate debate. Grimwade estimates that duties im- posed by the Community on cheap foreign imports, with the aim of protecting EEC manufacturers from what they see as un- fair competition, add around £13 to the av- erage price of a compact dise player, over £20 to the average price of a video recorder, £74 to the average price of ‘computer printers, £31 to the average price ‘of electronic typewriters and over £181 t0 the average price of photo copiers. Inall, these measures are costing UK ‘consumers and business over £274 million a year in the form of higher prices and EC ‘consumers as a whole £1,170 million. This is the equivalent of five per cent of total community spending on consumer elec- tronic goods. The Final cost to the public is even greater because businesses buying ‘equipment will pass on their increased costs to those who use their services. ‘Those are just the immediate costs. They take no account of the long-term effects, “If protection encourages inefficiency in the domestic industry, highers costs will lead to higher prices. Then, too, there is the deterrent effect that the mere threat of anti-dumping measures has on trade. Fi- nally, exporters to the EEC may raise their prices in anticipation of duties being im- posed. This latter effect may be even greater than the acwal rise in prices brought about by anti-dumping duties or price undertakings,” says Grimwade. Is this burden on consumers justified? The economic justification for imposing i LECTRONIGS such a burden on consumers is difficult to establish, says Grimwade. “The policy tight make sense if the competition posed by Far Eastern exports was in some sense ‘unfair’. This would be the case if it could bbe demonstrated that Far Eastern exporters were dumping their products at below their domestic costs of production, But in ‘none of the cases discussed in this report ‘was this clearly shown to be the case. Nei- ther GATT rules nor EEC regulations quire proof of predatory pricing. There is no necessity to show that export prices were below domestic costs to justify the imposition of anti-dumping duties. It is sufficient to prove that export prices were below domestic prices.” Even this has not been done fairly, argues Grimwade, who challenges the basis of the calculations. He points out that EEC manufacturers tend to argue that if sales of their products fall, this must be because of unfair tow pricing by competitors from outside the EEC. But this claim is often open to doubi, he says—the Japanese, for instance, suggested other reasons for the decline in EEC manufacturers’ sales of electronic ‘goods, such as their failure 10 update their technology, “Community manufacturers have themselves to blame for their loss of mar- ket share,” says Grimwade. “For example, they were slow in adapting their printers to make them compatible with IBM PCs when demand grew after 1983. The Japanese argued that they had increased their share of the market because their printers were superior in quality and per- Formnance.” In the case of photocopiers, the Japanese argued that the fall in EEC pro ducers’ market share was caused by bad investment decisions. Some companies did ‘not devote enought resources to develop- ing models that use plain-paper photo- copier technology. However, the EC largely rejected these arguments and im- posed! duties on all Japanese imports. Reduced consumer choice ‘As well as pushing up prices to consumers and business, duties on imported goods can affect the availability of a product if domestic manufacturers can not produce enought to fill the gap left by low-price imports. In some branches of consumer electronics, for example, EEC producers had already abandoned the cheaper end of the market by the time anti-dumping were imposed, Foreigh producers who specialize in cheap, mass-produced equipment may move up-market to avoid duty. “Far from sheltering Community producers from for- eign competition, the effect may be to in- tensify competition in another segment of the market. This may be the reverse of ‘what is required if the industry is to sur- vive,” says Grimwade. ‘Recommendations Grimwade recommends, among other things: + That before dutics are imposed on cheap imports from outside the Commu- nity, the cost to consumers should be cal- culated and weighed against the benefit to the Community's industry, There has never been a case where the EC has ruled out duties on the grounds that they would ‘harm consumers. + Anew, fairer definition of dumping: to ‘mean cases where an exporter sells abroad at less than_his production costs—not his domestic selling price. = Making it tougher to prove that the home industry has been injured by ‘dumped’ products. + A time limit on cach anti-dumping measure, + A single, international anti-dumping law. At present each country can write its own laws. + An intemational working group under the auspices of the General Agreement on ‘Tariffs and Trade—GATT—to implement anti-dumping policy. As national govern- ments would no longer decide dumping is- sues, anti-dumping policies could not be used as a weapon to restrict wade. Lady Wilcox, chairman of the National Consumer Council commented: “The pol- icy makers of Europe have listened for too Jong to only one side of the argument: that of producers in their own countries who fear that their own sales will be hit by cheap imports. Itis time that the voice of the consumer was heard, too." . ‘The National Consumer Couneil was set up by the Government in 1975 to represent the interests of users of goods and services ofall, kinds, The Council investigates consumers! needs and problems, reports on is findings and makes recommendations to government, Industry and others and then presses for changes in law and practice, where neces sary, in consumers! best interests National Consumer Counell, 20 Grosvenor Gardens, LONDON SWiW aDH, ‘Telephone O81 730 3469, Fax 081 730 0191 ELEKTOR, eCTRONI MAY 190,THE MULTI- MAT J. Buiting t CONCEPT ITT’s Digit-2000 system has been designed to ensure ready integration of new TV and audio standards with existing hardware concepts. This means that a MAC decoder based on the Digit-2000 system is readily installed into an existing TV set as an upgrade. Provided the necessary control software is available, it is, of course, also possible to use the relevant chip set in a stand-alone application, which is of particular interest to the many thousands of viewers who own satellite-TV receiving equipment. This article introduces the main components that go into the making of such a Intermetall/ITT Semiconductors is among the world’s largest producers of components for the consumer electronics market. Since 1985, the company has been involved in the development af MAC de coding systems, and it was the first semi. conductor manufacturer to introduce a D2-MAC decoder chip, the DMA2270. The planned use of D-MAC on the BSB ser vices, among other factors, prompted ITT Semiconductors to expand the DMA2270 ith a multi-MAC decoding feature. The result is the C-, D- and D2-MAC com- patible DMA2280, which, together with the DMA2285 MAC descrambler, forms the heart of the multicMAC decoder for the Digit-2000 system, C/D/D2-MAC compatible decoder. An important point must be made at this stage. When we speak of a MAC decoder, we mean a circuit capable of ex- tracting. video and audio information from a’signal to the MAC standard. As such, the function of the MAC decoder may be compared to that of, say, an FM decoder. Hence, the use of the word ‘decoder’ has in principle nothing to do with scrambling, and is really a mis nomer. Just like PAL TV signals, or, for that matter, FM radio signals, MAC ‘sig. nals may be encrypted. Since that process has basically nothing todo with the stand- ard of the transmission—only with the ‘way in which the input signal is pre-pro- cessed—a separate unit, the MAC de scrambler, may be used along with the MAC decoder. As already stated, the as sociated type numbers in this context are DMA2285 and DMA2280_ respectively. The use of the DMA2285is optional. How ever, bearing in mind that all BSB chan- nels are encrypted, a MAC descrambling chip like the DMA2285 is a must for all BSB receive units, Digit-2000: ready for the future The Digit-2000 concept is illustrated in Fig. 1. Signals travel from the left (signal sources) to the right (sound/ picture re ro tos itp creo Fig. 1. ‘an ADC and a DAC, ELEKTOR ELECTRONICS MAY 1990 ITT's Digit 2000 TViradio system is geared io easy expansion by virtue of a command bus and all-digital signal processing betweenj16] RADIO AND TELENISION, production devices). The intermediate signal processing is entirely digital be- tween an ADC and a DAC. Centrai signals for the system are conveyed via the IM bus, which isa simple4-wire network that enables a central or external processor to communicate with the various devices (slaves) connected to the bus, The system is very flexible in that itallows new stand ards to be implemented readily. Take, for instance, the MAC extension: I! is driven by the same ADC, is controlled by the same bus, and uses the same DAC as, say, the PAL cireuitry. This means that the sys tem allows both an economy and a top- quality TV set to be produced on the basis of three main building blocks: a fast ADC. 1 control bus, and a fast DAC. Extensions are always possible in this system: the appropriate unit (say, a NICAM proces: sor) is simply connected in paraliel with existing circuits and addressed via the IM bus. MAC in a nutshell The PAL, NTSC and SECAM colour TV systems currently in use are based on tre- quency division, which means that the two video components and the sound component are assigned a particular part of the transmitted spectrum. Jn this sys- tem, itis virtually impossible to ensare perfect separation. of the luminance Cbrightness’) and chrominance colour’) information. Inevitably, signals of both components will encroach upon each pther’s part of the trequeney: spectrum — see Fig. 2a, The effect is the well-known moiré patterning in picture areas with relatively fast luminance transitions. The colour processor in the TV receiver mis- takes these fast luminance signals in the cross-colour area between about 2.3 MHz and 35 MFiz (PAL) for colour informa. tion, and actuates colours which are not relaled to the luminance information in the particular picture area MAC relies on time division rather than frequency division and gives near perfect separation of the picture compo- nents, Figures 2b and 2c shows how the huminance (Y) picture components in a PAL video signal may be transferred and, compressed at a ratio of 3:2 intoa timeslot in the MAC signal (Ref. 1), The chromin- ance component (compression ratio; 3:1) is transferred in a similar manner to the time slot preceding the Y period. MAC lines alternately carry the compressed U (B-Y) and V (R-Y) colour difference sig- nals, Note that both Y and Ui/V are anal- ‘gue levels. Compression and expansion are required to fit-these signals into the available Tine time, which Is 64 ys just as ‘with PAL Bach line of MAC consists of serial U/YV and Y signals, reference periods and a sound / data burst (packet). The latter is digital and duobinary-encoded (Ref, 2) to reduce the bandwidth of the FM signal produced for D- and D2-MAC transmis- Sions via satellite. D-MAC differs from D2-MAC by ifs higher data rate in the inte Yow rae ee t nf ‘eet ST x 1 From PAL to MAC Fig. 2. sound /data burst: 20.25 MH». instead of 12.125 MHz (Fig. 2d), which allows a greater number of high-quality sound channels to be used at the expense of a slightly greater bandwidth Clock generator MCU2600 Aftera necessarily brief recap on the back ground of MAC, the components that go into ITT’s Multi MAC concept will be dis ‘cussed below with reference to block di- agrams, Unfortunately, the scope of this article does not allow a full description of teach device to be given; this may be found in the relevant datasheets ime-division multiplex of plelure components and sound. Time multiplexing must rely on accurate clocking of various circuits in the MAC decoder. As shown in Fig, 24, the system clock required for a MAC signal is deter- mined by the number of samples within the line time of 64 ys, and the line fre- quency: 1,296/'13,625 = 20.25 MHz "The MCU2600 supplies the digital pro- cessors, decoders, converters, etc., that form part of the Digit-2000 TV system with the required main clock signal, Which is of trapezoidal shape, with rounded corners, t0 avoid cross-talk and other interference. The MCU2600 may also be used for PAL, SECAM or NTSC: depending on the crystal used, the chip ELEKTOR ELECTRONICS: MAY 1990Fig, 3 Block diagram of the MCUZ600. supplies four times the chroma subcatrier frequency needed (PAL/SECAM: 17734 MHz; NISC: 14.318 MHZ). ‘All three VCOs on board the MCLI2600 may be selected individually, via the clock and data inputs, to form part of 2 PLL which is controlled by another chip, tor instance, the DMA2280 MAC decoder. The PLL control (= error-) signal is ap- plied in digital, serial, orm to pin 6 of the MCU2600. The default VCO selection is VEO. Video coder/decoder (codec) VCU2133 ‘The VCU2133 contains the ADC and the DAC mentioned above in the introduction ‘of the Digit-2000 concept. The chip is pro- vided with the baseband signal after de- ‘emphasis to the MAC standard (which Is not the same as the CCIR standard for PAL). As already noted, all digital signal processors in the Digit-2000 system are located between the ADC and the DAC in the VCLI2133, which provides the follow- ing functions ste Fig, 4) + twosottware-selectable input amplitiers: + onefast A-D converter forthe composite video signal + one noise inverter + one D-A converter for the luminance nal + two D-A converters for the colour dif- ference signals + one RGB matrix for converting the col bourdifference signals and thehuminance signals into RGB signals + three RGB output amplifiers + programmable auxiliary circuits for blanking, brightness adjustment, white balance control and picture tube align- ment + additional clamped RGB inputs for text teletext or other analogue RGB signals + programmable beam current clamping The VCU2133 may be used with a variety of video circuits, including the VPU2203, PAL, processor, the CVPU2233. NISC ‘Comb Filter Video Processor, the SPU2220 SECAM Chroma Processor, the DPU2553 Deflection Processor and\ the DTI2223 FLURTOR ELECTRONICS MAY 1990 Digital Transient Improvement Processor (note: DTI is sometimes referred as CTI: colour transient improvement). The chip contains 4 large number of registers that are loaded and read by the central proces sor in the Digit-2000 system via the IM bus, The A-Deonverter that follows the two video input amplifiers and the selection switch is of the flash type, which means that it isa circuit that consists of 2° com parators in parallel. For a slowly varying video signal, 8 bits are required. To achieve S-bit picture resolution with a 7- bit converter, a special operation known as ‘bitenlargement’ is used, During every other line, the reference voltage of the A-D canverter is changed by an amount corre- sponding to one half of the least-signiti- cant bit (LSB). In this manner, a grey value between two 7-bit steps is converted into the next lower value during one line, and Into the next higher value during the next line. The two grey values are averaged by the viewer's eye, producing the impress- jon of grey values with &-bit resolution Synchronously with the changing reter- tence voltage of the ADC, a hali-bit step is added to the output signal of the ¥ DAC every second line, The bit enlargement is switched off for D- and D2-MAC signals bby appropriate control of the registers in the VCU2I33, The ADC's sampling frequency sup- plied by the MCU2600 is 17.7. MHz (PAL/SECAM), 143. MHz (NTSC) or 20.25 MHz (MAC). The cunverter's resol- tution is 12 LSB of 8 bits. Its output signal is Gray-coded to eliminate spikes and glit- ches resulting from different comparator speeds, or from impertections in the coder sel After having been processed in other circuits, e.g., the DMA2280, the different parts of the digitized vieleo signal are fed. hhack to the VCU2133 for further process: ing to drive the RGB output amplifiers The luminance (Y-) signal is routed from the contrast multiplier in the DMA2280 to the Y DAC in the VCU2133 in the form of ABBREVIATIONS ADC —— Analogue-to-Digital Con- verter AGC Automatic Gain Control ALU Arithmetic Logic Unit BER —_Bit-Errar Rate BSB British Satellite Broad- casting CIR Comité Consultatit Inter- national de Radio. CCU Central Control Unit CLIMB Command Language for Intermetal! Bus DAC _Digital-to-Analogue Con- verter ™M Frequency Modulation HD-MAC High-Definition MAC MAC Multiplexed Analogue Components NICAM. Near-Instantaneous Companding Analogue Multiplex NTSC National Television Standards Committee PAL Phase Alternation Line Pc Personal Computer PDM_ Pulse Duration Modulation PLL Phase-Locked Loop QPSK —_ Quadrature Phase Shift Keying RGB ——_-Red-Green-Bine ROM Read-Only Memory SECAM —Séquentiel Couleur & Memoire SEEPROM Serial Electrically Erasable Programmable Read-Only Memory VCO Voltage-Controited Oscillator parallel 8-bit signal with a resolution of V2 LSB of 9 bits. This range provides ‘enough headroom tor large contrast vari roel ‘i Pe ||| eH fe Ee] ie Hee a [teu a owt me 900084412 Fig. 4 Block diagram of the VCU2T33 ADCIDAC.B RADIO AND TELEVISION ations as well as positive and negative overshoot: The Y DAC is an R-2R ladder network which is provided with the cen- tral clock frequency (20.25 MHz for MAC). ‘The (wo digital colour difference sig- nals, R-Y and B-Y, are transferred in a time-multiplex arrangement to save on input pins. Ata clock of 2025 MHz and a chrominance bandwidth of between 1 MHz and 2 MHz, this can be done with impunity. Like the Y DAC, the two 8-bit DACs for R-Y and B-Y are implemented as RR ladder networks. Although they are clocked at one quarter of the central clock frequency, the multiplex data trans- fer rate is 20.25 MHz (for MAC). Sixteen (four times four) bits are transferred se ‘quentially under the control of a sync sig- nal that co-ordinates the multiplex ‘operations between the VCU2133 and the video processor (in this case, a DMA2280). ‘C/D/D2-MAC decoder DMA2280 This chip forms the heart of the mulli- standard MAC decoder. Its tasks may be summarized as follows: * toaccept the digitized video (baseband) signal and extract from this the time- compressed chrominance and Iumin- ance information, and the sound/data packet + fo de-compress (expand) and correlate the luminance and chrominance infor mation + to extract audio, special data and syne ‘words from the sound /data packet, tak- Ing account of the two different data rates (D2-MAC: 10.125 MHz; D-MAC: 20.25 MHz) + toensure a central clock of 20.25 MHz by providing control voltage tothe PLL in the MCU2600 + recognition of packet 0 for special pur- poses + when required to provide error correc- tion on weak input signals, and allow different slicing levels to be defined for the on-board duobinary decoder + to provide an AGC signal for (digital) level control of thebaseband inputsignal ‘+ tocommunicate with the central IM bus processor ‘The DMA2280 is the multicMAC version of the (older) DMA2270, Its block diagram is given in Fig. 5. The DMA2280 is a com- plex chip by almost any standard because Ithandles many relatively fast digital sig- nals at the same time. It has on-board it minance and chrominance storage circuits which enable the relevant picture compo- nents to be de-compressed (expanded) and multiplexed (chrominance only) tunder the control of the central clock, Fu thermore, it is capable of de-interleaving. and linking the packets sent in each MAC. TV line. A special word recognizer with error correction capabilities ensures the recognition of the field and line synes, which are complex digital words con. Fig, 5. The heart of the decoder: the DMAZ280 mulil- MAC processor. tained in the sound/data packet The DMA2280 has a capability for direct inter facing with any of the teletext processors in the Digit-2000 series, such as the ‘TPU2233, The sound recovered from the data packets is fully decoded by the DMA2280 but left digital for demultiplexing and converting into analogue form by the AMU2485 audio processor. ‘All functions provided by the 1DMA2280 are controlled by registers, of Which the content is determined by the chip itself (read-only) or the central pro- cessor. The bit-error rate (BER) register, for instance, contains a number that rep- resents the sum of the error bits en- countered in the 82 packet headers in one frame. This sum is stored in a register that can be read as bits 0-7 at address 208 by the central processor, which can take the necessary actions such as muting the audio signal when the BER parameter ex- ceeds a certain predefined level. The DMA2280 occupies a total of 12 addresses on the IM bus. The bits reserved for these registers control a total of over 30 pro- grammable functions, some of which may be used to select, in turn, up to four diffo ‘ent modes of operation. The selection be- tween C, D- and D2-MAC is not automatic and must therefore be accom- plished by the control software, It should be noted that the DMA2280 requires a separate sound demodulator for C-MAG, since in that case the sound is provided in 2-4 QPSK rather than duobi- nary FM, Audio mixer AMU2485 The AMU2485 (Fig, 6) receives the serial audio data supplied by the DMA2280 at its S-bus inputs. The S-bus is unidirec- tional and consists of three lines; S-clock, S-ident and S-data. The sound informa- tion is transmitted in frames of 64 bits, divided into four successive 16-bit sam: Fig. 6 Block diagram of the audio component in the decoder, the AMUZS85. ELEKTOR ELECTRONICS MAY 1990THE MULTEMAC CONCEPT ® ‘cou 30xx Receiver ROM version PCeonirak 7 input Z. vo 2 Fan) (GRAM Baseband Baseband vou2133 DMA 2285 ener vouetss . i Processor tae serene ype mac oe « ent +0E- Emphasis Am ES, a Decoder pee ratp 8 ‘AGC YY Teuzran Teletext [sec | DRAPE st & se = [sae a st Fig. 7. Block diagram of a mulli-standard MAC decoder Intended for use as a sel-op box. The PC-lo-IMB interlace is required during the development stages only. ples. Each sample represents one audio channel. The repetition rate of thesamples is equal to the sampling rate of the D- or D2-MAC signal so that up to four sound channels can be transferred simultaneous- ly The AMU2485 provides a complen digital signal processor that runs its own control program from an on-chip mask programmable ROM. Two de-emphasis filters are available: one to the CCIR J17 standard (used for MAC and NICAM transmissions) and one to the 5D us stand- ard (used for PDM sound broadcasts) Both de-emphasis circuits operate digi tally and can be switched off if requires byan appropriate register instruction. ‘The oversampling filtersin $-bus chan nels 3 and 4 allow medium-quality D2 MAC sound signals (16 kHz sampling tate) to be mixed with high-quality signals (32 kHz sampling rate). The filters are third-order Cauer-type low-passes with 2 stop-band rejection of 40 db. The audio mixing feature of the AMU2SS allows any input to be routed toany output and, of course, to mis differ cently weighted input channels. The mix ing and volume control operations on the DACs are entirely digital and run under the control of an internal ALU that re- ceives the appropriate commands via the IM bus. A multi-MAC decoder The proviusly diseussed chips all go into the making af the MAC decoder shawn in ELERTOR ELECTRONICS MAY 1990 Fig. 7. This concept packs all the signal processing required between the base: band output of the indoor unit and the RGB drivers in the colour monitor or TV sel into a single set-top decoder. The IM bus, which has nat been dis- cussed sa far, is shown as a shaded path that links the sub-circuits into a small net work. The bus consists of three lines: Sig. nal Ident (ID), Clock (CL) and Data (D). The clock frequency range is 50 Hz to 170 kHz. Ident and clock are unidirec- tonal from the CCU to the slave devic data is bidirectional to allow the CCU 10, interrogate devices by loading and exam- ining the contents of their registers. The block diagram in Fig, 7 shows that the decader can be controlled either by a PC via the PC-todMB interface, or by a CCU which uses SEEPROM for storing and loading user settings suchas the MAC standard {C, D, or D2), saund selection or contrast. The PC is required only during, the development stages of the decoder: tho software that runs on it, CLIMB, allows all registers in the chips that form. the decoder to be examined and, it ne ary, loaded or reloaded. CLIMB allows individual chips such as the VCU2133 to be programmed in great detail, with the aim of developing machine code for the ecu, Once debugged and tested, the system control software is burned intoa ROM on board the CCU, The CCU, which may be 8 054 oF BOSA-like processor, has a direct input for digital data supplied by an intra red rece! As shown in the block diagram, the DMA2280 works in conjunction with the DMA2285 descrambler. In addition to its normal function asa low /high level MAC decryption processor, the DMA2285 allows 16:9 format HDMAC pictures to be converted to 43 tormat. Note, however, that this feature makes the decoder de- scribed only partly compatible of HD- MAC because of the present resolution of 625 lines, Fortunately, the nest generation ‘of MAC chips —which are now being de- veloped — will be capable of meeting the full” HD-MAC specification — with 1,250 lines and thus deliver virtually icker-tree wide-format pictures. Source: Datasheets AMU2485;__ DMA2280; DMA2285; VCU2I33; MCU2600/2632 DMA2270; CLIMB V2.1 ductors. ITT Semicon ITT Semiconductors UK. * Rosemount House © Rosemount Avenue * West By: fleot * Surrey KTIA 6LB. Telephone: (0932) 336119, Fay! (0932) 336145 Worldwide Mead office (TT Intermetall GmbH # P.O, Box 840 ¢ D-7800 Freiburg * West-Germany. Tele- phone: +49 761 517-0. References: 1. The MAC sy July / August 198; 2. Introduction to ry encod Ing/decoding. Elektor Electroniés January 1990. duotHORN LOUDSPEAKER 1. Giffard Horn loudspeakers for domestic hi-fi reproduction are generally large, cumbersome and expensive boxes that are built only by real enthusiasts. It is nevertheless highly satisfying to build them, particularly the one described here which, although producing excellent sound quality, is far less expensive than usual. Because of its high efficiency, it will produce more than enough sound pressure for most domestic purposes, even when it is driven by a low-power amplifier. The horn loudspeaker is almost certainly thealdest type of loudspeaker, dating back | Dfveunts ——_MeFarlow Type T860 to the days of Edison's wax cylinders, NeFalow Type H25:90 ‘Then, they were the arly devices to.offer | Yejynebl bot. about 300 les sound amplification, and. ever since | etincy Sie a searchers like Webster, Wilson, Voigt and Klikpsch have applied their skills to the 80 W continuous Power output sp ae a ee ‘eect In hii installations, however, the horn has had to give way long ago to different types of loudspeaker construction, such as the bass reflex, the transmission line and the closed box. Nevertholess, the domestic horn loud complex, while the drive units it uses are Be sive. Moreover, it takes up a lot of space in wrally (but not alsvays) quite expen the home Since it is very difficult to buy a horn speaker, the one described in this article has been designed specially for us by pro- fessional designers. The drive units and filter cost something like £50-£70 per box. The construction of the enclosure is, how: ever, nat recommended for beginners in speaker is not dead and for very good rea sons: it has a very high efficiency, and con sequently a lange dynamic range, good im pulse behaviour and low distortion. On the other hand, its design is usually highly woodworking The horn A horn is basically an former, It transf aphragm into an effective large area dl phragm without the disadvantages of i ass, cone resonances, and so on. di The radiation resistance of a lange-are: aphragm is much greater than that small area one and thus more power is ra lated for a given velocity of volume of ait (ound velocity) The basic requirements in the design a hom are maximum acoustic power, wide frequency range, and low distortion. Once these have been determined, the units may be specified throat and mouth diameters and the form and length of the horn may be calculated, Many horns are of the exponential after which the (others are conical or hyperbolic). The ex ponential behaviour of the horn ensures better coupling between drive unit and air ind this increases the efficiency to almost which is an enormons improvement OF Fig 1. Basie design of @ horn loudspeaker. Fig. 2. The MoFarlow wooter and tweeter. Fig. 3.Cicult diagram of the crossover filter: ELERTOR ELECTRONICS MAY 1990HORN LOUDSPEAKER o° Me Farlow 00087 14 ae nap te | ‘a0% fa eet - : - 20.0. | | Me Farlow ~ - Is 90008715, Fig. 5. Impedance characteristic of overal system Files L008 IN 1-23-98 1 its ie Tine - nsec aK Inguise Respons Fig. 6, The impulse behaviour ofthe overall system is typical of ahorn design, cover the 1% of most other types. The calculations of a horn design are not simple: every one of the many types has to be computed differently. We will not go into all of these, however, and will re- strict ourselves to a general description of the operation of a typical horn system, ‘The modern horn speaker consists of a ELEKTOR ELECTRONICS MAY 1990 drive unit and matching horn as shown in Fig, 1. The drive unit is loaded by the vol- ume of air in the compression chamber. Since the acoustic impedance is inversely proportional to the frequency, the com- pression chamber and the throat effectively forma low-pas ‘The throat provides the coupling be- filter tween the drive unit cum compression chamber and the horn. The area of the throat is important for optimum coupling The area of the mouth of the horn de- termines the low cut-off frequency. A good rule of thumb here is that the cireumfer- tence of the mouith must be at least equal to the wavelength of the lowest frequency to be reproduced, Depending on how many rfaces the horn will be coupled to, the ‘of the mouth may be reduced by a fac- tor 2 (floor), 4 (floor plus wall} oF 8 (corner of room). The length of the horn depends on a number of factors, particularly the ripple in the frequency curve that is acceptable. Since for good low-frequency reproduction the length is of the order of metres, the horn is normally folded a couple of times. Its total volume will then remain within acceptable limits for domestic use. Often, the drive unit is also loaded at the back by a horn or closed box so as to censure equal acoustic loading at both sides of the diaphragm. With backloaded horns, as used in the present design, that is rot possible, because these must then radi- ate the higher frequencies directly. SuchPry Aupio.s rt i — Fig. 7. Construction diagram of the horn enclosure. ELEKTOR ELECTRONICS MAY 1990Fig. : Artist's impression of partially completed enclosure. PARTS LIST (ONE ENCLOSURE) Drive units McFarlow T8-60 ‘McFarlow H25-90 Filter u alr-core inductor 06 mH wound fram 1 mm dia. ‘enamelled copper wire vB air-core inductor 0.25 mH wound trom 0.71 mm dia. e.cw. a 18 .uF, 35 V, bipolar c2 33 MKT RY 808, 5 W Re 202. 5W 18 mm chip board or plywood rear panel 250x112 mm top plate 250%420 mm. front panel 250x602 mm Rank (2x) = 4201150 mm panel 1 250186 mm pane! 2 250-326 mm panel 3 250x265 mm panel 4 250x159 mm pane! 5 250x385 mm panel 6 250x52 min panel 7 250160 mm panel & 250x186 mm pane! 9 250%68 mm panel 10 250x886 mm panel 11 250x100 mm panel 12 250.249 mm panel 13, 250.414 mm Miscellaneous. Damping material as required (ex: panded polystyrene nuggets or rack- wool) (glue and screws as required ELERTOR ELECTRONICS MAY 199 horns, including the present, are Uterefore used to reproduce the lowefrequency range aly Design parameters In the design of the present han speaker system the most important requirement ‘was that the mouth area should not exceed 0.125 m? so as to keep the dimensions of ‘the enclosure within reasonable limits. The throat area and the low cut-off frequency, for must also had to have reasonable val- ues. The throat area is normally given a value between 03 Ayand 1.0 Ay, where Ay is the effective cone area of the bass drive unit. Since the throat area and the volume of the compression cl acoustic load, and thus the bandwidth, of the speaker system, we have chosen a ratio of 0.73 (according to the calculations of W.M. Leach) [eis often thought that in horn systems only drive units with a very low Q), (that iti a very large magnet) may be used however, not always necessary: it depends on what bandwidth the system is required to reproduce. The bandwidth of a back-loaded horn as used in the present design isso small that a drive unit with a Qy, of 0.35%s perfectly suitable The low cut-off frequency is that fre- queney at which the horn is no longer loaded, that is, produces no sound. In the preseat design itis set at 40 Fz. The real =3 4B point lies somewhere between f_ and the frequency determined by the mouth area, The ratio of these two frequencies must not be too large to avoid irregular be haviour of the radiation impedance be- tween hom and room and a lumpy fre- quency characteristic The cross-sectional area, A, ofan expo ential hora at any distance x from the Uroat inereases according to the following equation: where Ay is the throat area, € 'm, the flare constant, = 2xf./ tHiesound Velocity (about 45 m/s), 718 and where is Drive units and filter In the choice of drive nnils it was impor tant, since the larger part of the frequency range is radiated direct by the drivers, to ind a combination that would match the efficiency of the horn,. The choice fell on te McFarlow Type T3-60 woofer and Type 125-90 tweeter (see Fig. 2) The woofer is a 20 cm type with a nor- nor LounsreakeR — RM mal pressed steel chassis and a coated paper cone. It has a reasonably sized mag nnet (dia, = 11 cm) and an efficiency of 92.dB (1 W/1 m). It is provided with a sep. arate aluminium front bezel that gives ita very attractive appearance. Moreover, it costs only about £30 or so, The tweeter has an even beiter eff ciency than the woofer, which makes some attenuation in the filter necessary. The domve is made from a type of pressed foam and the speech coil is cooled by ferro fuid Its price is very close to that of the woofer. The crossover filter, whose circuit is shown in Fig. 3, has been kept fairly sim: ple. The low-pass section has an attenua~ tion of 6 dB/octave and the high-pass sec tion one of 12 dB/octave. Because the crossaver point is rather high (3800 Hz), some impedance correction proved neces sary and this is provided by Rr and C). Re sistor R: ensures correct leve] matching be tween woofer and tweeter. The filter is eas ily constructed on a piece of veroboard or sf “ywood even on a small piece Building the encle‘e Most of the work goes into construc tion of the enclosure. You ca Feourse, have it made, but that may ix ase the cost of the system quite apprec. The construction plan is shown in Fig. hile Fig, 8 gives an artists impression 0, ar- tially completed box ‘The enclosure is made of 18 mmc. board or plywood: thicker board may used but st will then be necessary (and no, easy) to match the horn to the new dimen- sions. Wherever possible, angles have been kept to 45° or 90°. If possible, have the dealer you buy the board trom saw it to size according to the wood list. Start with gluing the rear panel, top plate, base plate, front panel and one of the side flanks together. Then, one by one, glue the inner wedges, inclines and tails in place. It is important to stick to the correct dlistances between all these panels. Any gaps where panels are glued together should be filled with a good-quality (sili- cone) wood filler. When the glue has set hard, drill holes for the connecting cables. At the same lume, fit the crossover filter in the hollow base Next, fit the drive units securely: in place , after which all the wiring should be completed, Finally, fll the enclasure with suitable expanded polystyrene chips or rockwool and glue the second side flank in place. The enclosure can then be finished ex- ternally to personal taste .PC SERVICING CARD Using a PC is one thing, getting it going again after a serious hardware malfunction quite another. The servicing card presented here, a design by ELV, is aimed at two groups of PC-XT/AT users: first, those bold enough to declare war on defective add-on boards, and second, those engaged in developing application-specific hardware. One remarkable feature of the servicing card is that it allows the card under test to be removed and inserted without the need of switching the computer on and off. sion slot on the nd foremost re moth without having to rebuild rvicing card is, therefore f the computer hat A further well-known source quirem urements with, say, an oscilloscope when that it extends the bus physically, s BASIC TEST PROGRAM PASCAL TEST PROGRAM Table 1. Example BASIC and Pascal programs that enable the servicing card to be controlled by the PC rather than push-button Tat ELERTOR ELECTRONICS MAY 1990peservieccarn Bg cru sol pl ; eo foe lieeey ee ees — fey O——— | wines Ost tore tomn_2P)s S Feset wee OFS ores LY vase = topints oe is sw 7 re peer O- sy} —Oiver | Ho at ot ce po" a but~ Signal Pin designation Signal | nae eee rae | va | Pare ss | [tee | gor | fae eB Boz noe | m7 | os | © | 00506 eo | 3 | ane os sos | 2 | Ae os sos| | ae oo | | aor |” |aor | oe | Frereves aos | [aoe] 01] ~12v_1 200 | ono | Bro | Caen | er asa at? ex ane [AIS AIT Ata | ANe | AIS mo | ate} 120 [AN aay aio, ne as} eel a ace | ar | ses | a8 | [woe | as | (eer [as ‘ss "nes | xe osc as} [ago | at [enor en] [ease [na Table 3. Signal assignment on the PC ex- pansion sia. ELEKTOR ELECTRONICS MAY 10 fers control signals OSC, ALE, AEN and CLK. Ina standard PC, the frequency of the OSC signal is 14 31818 MHz. The sys- tem clock, CLK, runs at 4.77 MHz since it is always one-third of the OSC frequency. Control line ALE (address latch enable) is, actuated at every bus cycle, and thusindi- cates that the CPU is not performing DMA access operation, A DMA cycle is indicated by a separate control line, AEM ‘The terminal count (TC) control line is buffered by three-state driver ICs: Circuit ICs buffers a number of control signals provided by the PC motherboard. Lines DACKO-DACK3 (DMA acknow- ledge) are driven by the DMA controller, which uses them fo issue DMA requests. Lines FORD and IOWR control read and write operations, Their equivalents for memory access operations are MEMRD- amd MEMWR respectively: ‘Control signals which are either bidi- rectional or supplied by OC (open-collec- tor) outputs are passed via reed relay contacts. This arrangement obviates the need of complex address decoders and direction control circuits. Datalines DO- D7 are passed via reed contacts REi-REs, and control lines [OCHRDY and 1/OCHK via reed contacts REs and REV, The [/O. channel check (I/OCHCK) line serves 10 signal parity errors in external memory areas. Such errors generate 4 non-mask- able interrupt (NMD. The IJOCHRDY (1/0 channel ready) line enabies bus cycles to be delayed. This is particularly useful for relatively slow. input/output ports or memories which require the bus access time to be Teng. thened. Control line CARD SELECT is passed via reed relay contact RE. Relay contacts RE}1-REys pass the sup- ply voltages, +5 V, -5 V, +12 V and ~12V to the card under test ‘The only fixed connection between the PC and the add-on board under testis the ground line, This ensures the presence of 8 reference potential the instant the acld ‘on board is inserted, and prevents open: collector outputs being damaged. ‘One of interrupt request lines 1RQ2~ IRQ? is passed via relay contact RE and wire jumpers Bri and Bri, which are fitted in accordance with the IRQline used, This ‘enables current to flow from, say, RBs via BRs, and on via REinand Br) to Bss, oF from RB2i to B4, ete. The type af add-on board PC SERVICING CARD NO Address Function (000-00F" DMA-Controliar (82378-5) 0201-021" Interrupt-Controller (#259-5) 0401-043 TimerCounter (8258.5) 0604-0534 System Register (825545) (B01-083H OMA-Page Register (74L$670) DADH-OBFH — NMI-lnterrupt Register DCOK-OFFH Reserved 1OO1FFH Front Panel Controller 200H-20F4 For Computer Games (Game Port) 210H-217H Additional Unit 2201-24FH Reserved 278H-27FH Second Printer 2F8u-2FFW Second Serial Interface 200-31F Prototyping Card 320H-32Fu Hard Disk-Controller 378H-37FH Printer Intortace (parallel) 380-38FH — SDLC-Inlerlace BAGN-BAFH Reserved 3B0K-98FW —Manachrome Adaptor and printer COH-3OFH Reserved 3D0H-3DFH Colour Graphics Caro SEOH-3ETH Reserved 3FON-3F7H Floppy Conirailer SFB\-GFFH Serial interlace Table 4. PCO address map and reserved functions (all addresses in hexadecimal). tobe debugged determines which jumper is to be installed. Line IRQS is commonly tused by the serial port, line IRQS by the floppy controller, and IRQ7 by the paral- Tel port. This leaves IRQ2, 18Q3 and 1RQS free for special applications and future ‘extensions, The use of the DMA request bines also differs from card to card, DRQT has the highest priority, DRQS the lowest ‘The reset line formsa special case. Nor~ mally, itis buffered by three-state driver ICs, When the servicing card is actuated it automatically generates a short reset pulse for the add-on board. i 9504 13 Fig. 3. Construction details of the cover plate which Is attached to the circuit board.OCESSORS. | COMPUTERS AND MICKOP eens conan spp by ELV Reser coo ae o + 2a me s 1 Ko m e 2 4KT a 2 iz 1 RoR Rie: Tax | Copectr: Tone 7 8 te bn * |2 tors 3 f | sanioncctoe: 32 rrat300 rs oe 1 7aLs08 ion amr 1 74.514 ‘cs me 1 fausse ion on 1 fase ioe 1} taus03 ee + asia ic 1 pastas eo 1 fatsese (orcs + pases ‘or 1 Cosoeo Ice 2 bosta ne 2 no: bie ; LEDSimed Misael 18 toad rcey eRe ? petimentputnuton Tar 1 Gena PC oxpanson sok 1 pniotultooe = 1 100.0m flexible wire 206m siivered wire 1 PC card cover pate Control circuit This consists of a 4-bit binary counter Type 7ALS93, [Cw, and a binary-to-deci mal decoder Type 74.5138, ICs. The oscil lator in IC, a Type CD4060, is set to operate at about 10 KHz. Its Q10 output therefore supplies a clock signal of about 10 Hz. This signal 1s applied to the clock inpot of counter ICin via Irs and IC When the counter reaches either state or t f t Sp eaa te, the eulpot of AND gate [erie goes slelelelielel e NAND gate IC blocks the clock pulses FPePE Sy ey Sys tobe provided with aclack plac via I 2. |P o-j-o}] el} e-ff-0 |} © Tits and 1Cr. This causes the output of elle le) 2 IChu ta change rom gw te Highton Seared 6 ling divider es ; ¢ A cloce pulse mayiniso be pfovidedioy 2 ffefe |] o}-of-e an 1/0 read operation on pat of the FC Baal sil 3] 2]. Fonter details ‘on this are given im the e [effelel elope: & state. is reached, the counter is reset to 0 elle lolol] ojo via 1G ana ICoa. Again, this can alto be accomplished by a PC I/O write oper svrtthed on the rset pulse causes bute Fig. 4, Component overlay on the Gouble-cided trough plated printed ove beara. ‘by pulleupresistor Rat pin 3. Ths causes ELEKTOR ELECTRONS MAY 1990the servicing card tobe disabled when the PC is switched on. The second part of the control circuitry decodes the individual counter states, Thicisdoneto ensure welltotined wn /oit switching of the address-, data-, control- and supply- lines. Reset line ST) is only factuated at counter state 3 and thus pro- vides a reset pulse via ICs. This pulse resets the servicing card toa default state The relays that pass the data signals are actuated during counter states 3and 4 via IGits, [Ce and Ri-Ts, while the control and address-lines are passed at counter states 2-5 via gates [Cr [Che and IC) In addition to these four states, the supply voltage is also actuated during state 1, The function and operation of the counter are summarized in Table 2 WO address decoder ‘The third part of the cireuit is formed by the 1/0 address decoder. The servicing card occupies oneaddress only in the PC's 1/0 map. AU contro! functions are derived trom the PC accessing this 1/0 address, Le, data, whether read or wnit- ten, is ignore ‘The 1/0 addressing function js based fon a lW-bit decoder which is set to the desired address with the aid of wire jum- pers Br=-BRis. This arrangement enables the servicing card to be selected only when the I/O address supplied by the PC is equal to the address set with the wire jumpers. The outputs of NOR gates 1Cix And ICjss are low simulianeously when the levels on PC adress fines A and AB are equal to the levels set with Bro and BR\s respectively. Only in this condition dloes the output of OR gate [Cos supply a low level Circuit [Cuan 8-bit comparator, is en- abled by a fove level on the AEN line. Its ‘output goes low if the levels of address bits AO-A7 mate those provided by wire jumpers Brs-Bri. PC bus line JOR is low when the CPL performs a read operation. This low level tables gate [Cir so that counter ICw is provided witha clock pulse, Ifthe write Tine, {OW, is actuated, the output of OR gate IC\w goes low and causes IC to be reset t0.0000 Construction ‘The complete circuit is accommodated on a double-sided, through-plated printed Circuit board of dimensions 2735104 mm. Two angied pieces of aluminium are used tosecure the card to the usual cover plate required for PC add-on boards. Cutting and drilling details of this plate are given in Fig. 3. ‘The construction of the servicing card is straightforward. IC sockets are not used 50.a8 to eliminate the tisk of bad contacts. All parts must be fitted at the lowest possible height to prevent them fouching parts on an adjacent board installed in the PC. Use precision pliers to press pin pairs in the harizontal rows of the 62-way bus expansion socket a litte closer togetiwerso FLEKTOR ELECFRONICS MAY 1990 that they can be soldered direct to the ‘copper tracks at the long side of the board fas shawn in the photograph. Use two M3xty nim screws and nuts to sectire the support bracket on the board 10 the cover plate. Address selection ‘The input/output address range and the associated funetions used in 4 standard PC are given in Table4 The 1/O address occupied by the ser- vieing card is set by wire jumpers Br Bris. As an example, suppose theservicing, card is to be controlled via 1/0 address 30. The tirst digit, 3, is equal to 112 and therefore set by leaving Brvi and Bris open. Since theall the other digits are 0, Bro=Bei> are closed, Debugging Although problems with the practical use Of thw servicing card should be rare if the construction is earried out with care and precision, a few hints are given to assist in saultrinding, Incase theservicing card causes the PC to crash, remove i from the expansion slot. Connect the card to an external 3-V power supply, The + is connected to bus contact B19 oF BAS, and geouind to BOI, BIN for BST, Measure the current consumption: this should be between 200 mA and 300 mA, Actuate push-button Ta) to check that LED Ds lights after a delay of about 05 5. If this sloes not happen, take a hard look at oseillator /counter ICs. This is normally disabled via pin 12. Pressing the button should provide IC witha clock pulse. As a result, pin Fof [Ci should go high 30 that the counter starts fo count either (0 state (or state 4 at which it disables IC: When the counter is not actuated, the btt combination at pits 1, 2and 3 uF I» is either at 000 or 100, The states of outputs ST1-STy may be checked with reference to, Table 2 In the actuated state, thee current con- sumption of the servicing card rises by about 70 mA. The current consumption then lies between 290 mA (typical) and 360) mA (maximum), The operation of the 1/O address decoder may be checked manually in case the relevant circuitry does not respond to the test program, The card adlress is as sumed to be 300 as in the above example. ‘This means that wire jumpers Bry-Brisare installed. ‘Connect inputs RA24-RA31 to ground. in RATL {AEN line} must be made Low in any case. Address lines AS (RA23) and AY (RAZ) are tied to +3 V. Check that pin 1 ff the 9-bit comparator, (Ci, is low. All ‘other inputs must be low as well. As a result, pin 19 is low also, A brief low level at RBIS (OWR) should cause the servic~ ing card to switch to its default state Make RBIS high again and briefly actuate RBI (fORD) by making i low. Apply a otyple of puilses in this way and cheek that they enable and disable the card via Ii A complete kit of parts for the PC servicing card is available from the designers’ exclusive worldwide cis ‘ibutors (regrettably not in the USA FRANCE Telephone: +33 82837219 Fax: +33 82838160 Aliso see ELV France's advertisement elsewhere in this issue, and IC If the service card does not ftinetion coneetly alter inserting it into the PC, itis recommended to remove itand first check the relay contsets RE\-REMafter applying 5, This is simple te achieve by measur ing the contact resistance between the relevant points at the both sides of the hoard, near the e\pansion bus socket, The ‘on’ resistance ol the contacts should be of the onder of a Few alims. For this test, ibs necessary ta select the service card and Switch on the extension funetion, Evident- Iv, jumpers Bri-Bry must be closed to be able to check thal the IRQ and DRQ lines are connected by the relevant relay con tacts. The correct operation of bus drivers IC)-ICs may be verified by applying high ‘and low levels to the relevant control pins at the reverse side af the board. Provided. the card is actuated, measurements may bbe made at the expansion slot Finally, the service card may be left in its PC slot even it is not used. .AN EXPERIMENTAL ALL-WAVEBAND FERRITE ROD ANTENNA Richard Q. Marris, G2BZQ The ferrite rod antenna described here is a most unusual conception in as much as it covers the frequency spectrum from 125 kHz to 24 MHz (2400-12.5 metres) with continuous tuning and without wavechange switching. This is in sharp contrast to the ferrite rod assemblies usually found in MW/LW radios The unit was designed as an an tenna for use with any all-waveband radio with external antenna requirements such as communications receivers, vin home constructed radios and HF inds of modern transistor radios. A fur ther and novel feature is that it covers the segment between the LW and MW bands, which is now have built oF pur W) and 2400 me- activity. cies, and tl tres where th 1 by either coaxial cable or 300 2 flat twin feeder, which is of particular interest Where end-fed long-wire ant be attached. In the case of older commuani- cations and domestic receivers, it wa unusual for a 100 f¢ long outside wire a tenna to be specified. Even if you have the real estate necessary for this, it isnot going to be popular with the neighbours and local authorities Many. successful experim: been conducted over the enti covered by the unit. The impressive n are in no small measure due to th rot selection of ferrite core material grades and sizes and the use of a 3-gang variable tun ing capacitor, The experimental unit is seen from var- ious angles in the photographs, in one of which (4) it is shown on top of a superb vintage Pye 9-waveband all-wave Export Receiver. This receiver is in everyday do: mestic use: its audio reproduction outper forms most modern AM radios. Since the unit is experimental, certain imperfections in the photographs owing to modifications carried out during the final construction, testing and evaluation, but 2 can be ELEKTOR ELECTRONICS MAY 199)000 in Taare RX |e ewe | é 4 | rtrescasasany | pe 130, z a 4 | = | fees secpaatemnm i\ retention 8 Store ELEKTOR ELECTRONS MAY 1990 that's experimental work for you! The circuit ‘The circuit diagram in Fig. 5 shows five tuned circuits: Li-C1; LC2; LCC Li-Chyand L-C5-Cr. A 5-gang, 500 pF per section, variable capacitor is used for Ci to €5, The location of coils L2, Ls and pling the antenna to the receiver, is critical All inductors are wound on ferrite rods. ‘The tuned circuits are adjasted simulta neously and cover different wavebands with small overlaps. The required band is selected automatically by the tuned input circuits in the receiver. There is no interac tion etween the five tuned circuits during, normal operation. The prototype has been evaluated and tested with several types of communica tions, domestic and vintage receiver: The wavebands covered by each of the five tuned circuits are Lis) — 125-450 kHz (2400-667 m) L3-C2— 400-1900 kHz, (750-158 m) L-CHCs— 1000-4250 kHe (300-71 m) [eC — 3500-9500 KHz (857-316 md LAC5-Cr— 8000-24000 KE (375-125 m) Construction The baseboard assembly shown in Fig. 8 is made of two pieces of 18-22 mm thick (ply}wood. The vertically mounted cop- per-clad board is for direct common earth- ing connections and is trapped between the two halves of the baseboard and se- cured with two brass angle brackets. All wood parts should be given an application ‘of teak colour wood dye. The LEMF coil assembly—see Fig, 6 — uses a 1309.5 mm Grade F14 ferrite rod, cut down from a standard 140 mm long rad with a Junior hacksaw. Grade FI4 is.a nnickel-zine material that is usable up (© 2 MH2, where performance just begins to {all off, The rod is clamped at either end in 2 plastic, round cable clip that is secured to a vertical bar of perspex (hardwood would do). The assembly Is bolted to the vertical copper-clad board as shown in Fig. 9, side view ‘A’ Both Lr and La are proprietary induc- tors; each has a small coupling winding Which should be ignored. The coupling coil to the receiver, Lz, consists of 22 close wound tums of 32 SWG enamel copper ‘wire ona small paxolin former. All three should be positioned in the exact positions shown in Fig, 6: deviations will change the overall frequency coverages The HF coil assembly—see Fig. 7—is also mounted on two vertical perspex (orcopper-clad board 2in. dia. cord drum ‘wooden base knob (with cord drum and slow-motion drive behind) drive cord with tension spring brass a bracket slow-motion drive shaft coupler ‘9= feeder terminal block ———— PARTS LI FRI = ferrite rod; Grade FIA: 140%9.S mm, cut to 130mm: Type FRA stock no, 35-14 147; Circkit Distribution Lid errite rod; Grade 61 7.50.5 in; Type RG 1-050-750; Amidon Associates, 12033 Oise. zo St, Nth Hollywood California 91607, US. L1 =amenna coil LWC1 stock no.35-00108: Cirkit Distribution Lid 13 =antenna coil MWC2 stock no. 35-0268: Cirekit Distribution Lid c1,c2 large BC type variable capacitor: J. Birkett €3,C4, C5) 500 pF large BC type variable FR2 << au sang x 300 pF capacitor; J. Birkett 470 pP, silver mica or 1000 pF, silver mica or | | slow-motion dri RX42V; Mapli shaft coupler (2 off) =0.25 in. shaft; J, Birkett dial cord drum (2 off) with ode spring; 54.5 mm dia, code RX43W; Maplin copper-clad circuit board, un drilled; 20090 mm 5 cable clip. plastic (4 off) tt | | On ferrite rods | wire (a) 32 SWG enamel insulated copper (b) 1/0.6 mm single strand PVC covered, 1.2 mm outside diame. ter (NOT 1.0 mm O/D!) ELERTOR ELECTRONICS MAY 1990fharstwood) columns that are later screwed to the vertical coil support posts shown in Fig. 8. The ferrite rod is 7.5505 inch (the US manufacturer specifies the dimensions only in inches). It is made of Grade 61 nickel-zine material, specified for use up to 15 MFlz, but which here, with spectal winding techniques, has been stretched to ‘over 40 MHz. “The coils are closewound straight on to the rod from single-strand 1/0.6 mm PYC covered wire, 12 mm overall diameter Similar wire exists with thinner insulation ‘of about 1.0 mm overall diameter, but this MUST NOT be used. The direction of winding should be same for all five coils They should be positioned exactly. as shown m Fig. 7: deviations will result én. incorrect frequency ranges and incorrect coupling. After the coils have been tested, they should be held in place with spots of adhesive at the ends. The number of turns for the oils is 36; Gs le —15 sang variable capacitor is made by coupling a 2-gang, and a 3-gang capaci- tor together with the aid of two cord-drive drums, a length of dial cord and a spring for tension as shown in Fig-9 and the pho- tographs. A single epicyclic slow-motion rive operates the two variable capacitors simultaneously. The variable capacitors are of the large broadcast’ type that have the advantages ‘of wide-spaced plates, ceramic insulation and a full 500 pF maximum capacttance, while their large size assists in keeping the leads to the coils short. Moreover, the mini= ‘mum capacitance is smaller than that of ‘most smaller compact types. In view af possible variations in the di- mensions of different makes of capacitor, the exact mounting positions are not shown The assembly should be fitled as shown with the aie of shaft couplers and lengths bf 0.25 in. dia. rod to align the assembly and drive mechanism. On the prototype, the 2-gang variable capacitor was screwed. vertically to the baseboard with the metal frame hardwired to the vertical copper clad board. The 3-gang was mounted up- side down (to reduce lead lengths) and bolted t0 the copper-clad board, All earth connections should be sol- dered direct to the copper-ciad board by the shortest possible route. Coupling coil L2 is connected to the ELERTOR ELECTRONICS MAY 1990 AN EXPERIMENTAL ALL-AWAVEBAND FERRITE ROD ANTENNA, feeder terminal block—see Fig. 9 side view ‘B’—by a short length of 300 0 twin feeder. Coils Ls and ries and also connected to the terminal block via some 30) 2 twin feeder and then wited in series with the feeder from La. ‘This means that all three coils are in series, as show in the circuit diagram. ‘The feeder to the receiver may be either the ustal coaxial cable or 300 0 flat twin feedline. This gives versatility of connec tion {all types of receiver input imped- UW night be thought that the coupling coils should have dissimilar numbers of turns, depending on the feedline imped- ance. Practical experiments indicated that only a fraction of a turn difference would be necessary on L5 and 17, while La was not critical. Because of that, the numbers of coupling turns are a compromise that does not degrade the performance, however: Testing ‘The correct feeder should he connected be- tween the feeder terminal block and the receiver. [twill be found that the tuning. of the antenna tnit is quite sharp, thus im- proving the selectivity of the ree Whatever the selected waveband, the an- enna tuning should be brought to reso nance as indicated by a significant inerease in signal. The antenna is direetional, with ‘maximum signal appearing om the “flat” side of the rod and minimom signal at rhe ‘ends of the rod. Rotation of the unit will, therefore, increase/ decrease the strength. Of the receives! signal. If there is interter- tence from other stations, local man-made noise, or static, Ure antenna should be ro~ tated slightly to either side to reduce /elint- inate that interference. In general, it will be noticed that the ambient noise Tevel is far lower than with a long-wire It will have been noted that the earlier quoted frequency ranges have small over- laps, so that the whole range of 125 kHz to 24 MH is covered by five complete sweeps of the S:gang capacitor to match any preselected receiver frequency range whether LF, MF or HE A preamplifier may. be needed between the antenna and the re~ ceiver in the higher HF ranges 1f the re ceiver does not have a high RF gain The exact frequency ranges can be checked, if required, with a signal genera~ torand appropriate receiver. If as is prob ble, a signal generator is not available, but the receiver is dial calibrated, it ts possible to check the ranges with an artificial noise signal. For this, a small battery-operated caleulatar is placed about 12-18 in. from the antenna. This generates.a noise signal that can be resonated by the antenna-re- ceiver combination. By manipulation of the calibrated receiver tuning and the an: tenna tuning, itis possible to check the fro quency ranges of the five antenna ranges and, most importantly, to check that the ranges overlap somewhat to provide con: tinuous frequency coverage tena, Conclusion This compact antenna unit covers all those frequencies, HF, MF and LF, that one may like te: receive below 24 MHz. tt is direc tional tor elimination or reduction of adja- cent station interference, man-made elec Inical noises and static. It is far smaller than conventional antennas and picks up for less noise. It could be housed in a simple polished or painted wood enclosure or plastic (not metal case, .INTERMEDIATE PROJECT A series of projects for the not-so-experienced constructor. Although each article will describe in detail the operation, use, construction and, where relevant, the underlying theory of the project, constructors will, none the less, require an elementary knowledge of electronic engineering. Each project in the series will be based on inexpensive and commonly available parts. ACOUSTIC TEMPERATURE MONITOR J. Ruffell Electronic temperature monitoring need not be complex. This circuit gives an audible indication when a preset temperature is reached, It can be used as a thermal alarm on boilers and heating systems, as a bath water temperature indicator, or in an electronic Basically, the temperature alarm consists of three parts: a temperature sensor, 3 small printed-circuit board which con- tains the measurement circuit, and 3 piezo-ceramic buzzer that functions as an Indicator. The printed-circuit board and the buzzer are housed in a small ABS en- closure, together with the battery that powers the circuit. The sensor is con: nected to the circuit via two lengths of Hexible wire so that it can be located as close as possible to the abject whose ter perature is to be measured Temperature sensor The temperature sensor is formed by an integrated circuit rather than the more conventional NTC (neg mperature coetficient) resistor. This choice was made to ensure a wide temperature range with acceptable accuracy. The sensor used Type LM235. This device, which looks like an ordinary transistor in a plastic enclo sure, contains a fairly comples circu! that provides an output voltage that isa func tion of ambient temperature. The output age is linear over a wide range and rately defined at 10 mV/K. Since the electrical behaviour of the LM235 1s not unlike that of a. temperature-sensitive zener diode, a similar crcsit symbol is used. A third pin is drawn, however, to indicate that the LM235 has a calibration Input (which is not used here). The sensor is supplied in a number of versions for different temperature ranges, The LMI35 is suitable for -35°C. to +150 °C, the LM235 for ~40 °C to #125°C, and the LM335 for ~40°C to +100 °C. In principle, all three may be used in this whistle kettle. circuit: the choice depends on the applica tion. Note, however, that the price of the dovices rises with the temperature range. The circuit The circuit diagram i€ given in Fig, 1 Apart from the sensor, only one mte- grated citcult is used. This IC, a TLC contains two CMOS operational ampli: fiers, At and Az, The first is used for the lwmperature measurement, the second for the acoustic alarm, Opamp Ai is wired as a comparator: i compares the voltage provided by the temperature sensor, Ds, to that provided by anv adjustable reference, Di, The LM336 in that position is a temperature-comper: sated 5-V zener diode. The reference volt age may be adjusted between 0 V and +5 V wilh the aid of preset P) before itis applied to the +input of comparator A\ ‘As long as the temperature measured by Deis relatively low, the voltage at the ‘input of A) is lower than the reference lage at the +inpul. Asa result, the opt unser to 1%) auazseteneere Fig. 1 Circuit diagram of the acoustic temperature monitor. ELEKTOR ELECTRONICS MAY 1990Fig. 2 Suggested construction on univer- | prototyping board OMPONENTS LIST Resistors: Rr Re Fe eRe Fe Bs Pe geo ii 1. push-o-make bution Ss 1 miiatueoniefswicn Se 1 Ba 1 9V battery Bs 4 PBS put of the comparatpr Is high. Le. vite tually equal to the battery voltage. The ‘oscillator, Az, isdisabled because the june equency-determining compo- is held at about +9 V via ELEKTOR ELECTRONICS MAY 1990 ACOUSTIC TEMPERATURE MONITOR Fig. 3. Completed printed-cireult board with external components connected. diode Ds. Hence, buzzer Bz) remains si- lent This condition is ended when the measured temperature rises above the set threshold. In electronic terms, this means that the voltage supplied by Ds is higher than that at the input, so that Ar toggles and supplies a low output voltage: Diode Dy blocks and decouples the output of Ar from the oscillator, As. Buzzer Bz) is actu- ated and supplies an acoustic signal of which the frequency is determined by Ps, Push-button Si allows the cireuit to be reset following an alarm condition. When pressed, it causes Ci to be charged to the supply voltage, so that the voltage at the sinput oF Av is higher than that at the input, irrespective of the voltage sup: plied by Ds, Pressing Si therefore disables the oscillator, Evidently, C) will be dis- charged slowly via Po and R). This takes a while, however, becatise of the relatively high vale of the components. The upshot is that Ds will have cooled down to a tem- perature below the alarm level well before the voltage at the +input of At has fallen below the reference voltage. Construction and adjustment A suggested arrangement of the compo- nents on universal. prototyping. board size-1 (UPBS-1) is shown in Fig, 2. The population of this PCB should not present ny problems. The buzzer, the tempera wre Sensor, the battery and the two swit ches (reset and on/off) are external components, which are connected to the circuit via wires and solder terminals. In most cases, the alarm will be set for a fixed temperature, e.g., approximately 100°C for botling water. This allows a preset to be used as shown in the component mounting plan. Ita variable temperature setting is required, P) is replaced by a potentiometer which is fitted on the front panel of the enclosure ‘The way in which the sensor mounted and connected to the eireuit de- pends an the application, For temperature ‘measurements on fixed objects, the sensor is simply secures! with a small clamp or a plastic cable tiv. For measurements on hot gases, mount the sensor at the end of a probe and insulate its terminals with shrink Sleeving or two-component epoxy ‘The alarm is simple to adjust with the aid of a thermometer. Immerse the sensor (not its terminals) in water which is heated until the desired alarm tempera lure is reached. Wait a few seconds, and adjust Pi until the buazer just starts to sound, Next, set the frequency of the alarm tone by adjusting Ps In cases where the alarm is used con- tinuously, as with a CH boiler, the battery ‘may be replaced by a mains adaptor with 9-VDC output. The current requirement of the alarm is modest at a few milli-amps only, so that a low-power adaptor may be used .CENTRONICS A-D/D-A CONVERTER J. Ruffell To the many PC users who would like to interface their computers with the real world we present an analogue-to-digital and digital-to-analogue converter. The low-cost, versatile, unit with accompanying control software is unconventional in that it is connected to the PC’s Centronics port, which is normally used for a parallel printer. The use of stanclard interfaces Forapplications they are not in tended for is widespread and goes back to the days of the first hobby computers, The ad intages are obvious: there is no need to get fo grips with the atdware, and the function of the peripheral is not dependent on extension connectors peculiar to the sys: tem. ‘Thus quired — to custom-made’ the software re control the peripheral is often hardware-dependent and ol by reweiting the systenv-resident 1/0 routines, accessing the relevant cir cuitry in. non-standard way eg, through bypassing. the BIOS (baste input/output sys Many modern PCs contain large gate arrays instead of in dividual 1/0 circutts. The A= D/D-A. converter described ed without the need writing the control rou: ines available for parallel |/O operations. Note hat this does not imply complete inde. dence of the hardware, since the BIOS routines normally tased for controlling the Centronics printer port are nat suitable for the controlling the present converter however, rd. Fortunately, the degree of hard: fe dependency is Jow and restricted to 1/0 routines, The con: trol program available for the converter board should ne give prob- lems on most MS-DOS computers. Note, however, thata number of older PCs have & printer port with incomplete handshak Ing. The absence of certain lines generally does not cause problems when a standard printeris used. The converter board, how ver, may require these lines fora number Centronics port inputs The block diagram of the A-D/D-A con ELEKTOR ELECTRONICS MAY 1990 verter is given in Fig. 1. The operation of the circuit is based on the use of the output as well the input lines of the Centronics port. The latter are nor vey ‘paper empty’, ‘busy’ and other infor imation from the printer to the computer ‘The converter, however, rises these inputs, to conyey digital data, such as the state ot two comparators, to the computer, The ally used to con two comparators enable two analogue input voltages to be compared with an analogue output voltage supplied by the DAC (digital-to-analogue converter]. By writing a series of rising values to the DAC and monttoring the relevant con parator output, the computer is able to determine the value of the analogue input voltage applied to the board. Such an operation is generally referred 1o as suc cessive approximation, In the present advantage lies in the use of a single computer input only instead of a number equal to the conversion resolu tion in bits (in this case, eight This is an important consider ation since there are few inputs fon a Centronics interface. A LED monitorcircuit enables, by selection, either the state of the Centronics data lines or those of the digital outputs of the circuit to be indicated. Circuit description The actual circuit (see Pig. 2) is just as straightforward as the block dia ram, although some details may create an impress ion of greater complexity, The ight databits on the Centronics portare fed direct to the inputs bf the DAC, [Cia Provided the PC port meets the drive specifi cations set out tm Centronics standard, these lines are driven (OC) outputs, by open-collect MAIN SPECIFICATIONS ‘BIE D-A converter = output voltage: -6 V10 +5. ~ total setting ime: approx. 1 ps + three reference options: REF-02 (+5 V; vely stable) TLS17 [<5 V; low-cost) external source 2-channel A-D converter = DAC:-basad successive approximation teruators for adjustable input sens tity 44 multi-purpose OC outputs + legay=100 A; Ucn 20 V 3,4 or 5 digital inputs ~ swiiching threshold: approx. 25 V + CMOS and TTL compatible LED indication tor functional checks ~ monitors either Centronics datalines ‘ data inputioutput inesCOMPUTERS AND MICROPROCESSORS ‘ethe use of pull-up network Ry, The contput current supplied by the DAC is determined by the value of the dataword supplied by the computer. This current is converted into a voltage with a range of 5 V to 45 V with the-aid of ICs and ICi0. ‘The relation between the applied data and the output voltage, Li, is expressed by Uy = Une (data-128)/128 LVI The reference voltage, Use is supplied jumper jumper Le 6 #H oy oo Do Da ot D1 be ca De Dro. 0 03 on i De Die 2 05 Dis 3 Ds ov wat o7 Dis. Isial2 are ‘Table 1, LED Tunction overview, Block diagram of the Genironics-compalible A-D/D-A card. either by an external source (via jumper B), or by an internal source. The internal source is either a REF-02 or a TL317; the REF-02 provides the highest stability and accuracy, but is more expensive than the L317. The choice between these two de vices is up to you. Opamps [Cis and [Crys form the pre viously mentioned comparators. Their in- puts are protected against overvoltage by two diodes (D»-Di/Ds-Ds) and a resistor (Reo/Rw). The input sensitivity may be adapted by modifying attenuators Re-Rex and Ru-Rio. The indicated resistor values provide an attenuation of 2 times, which creates an input voltage range of 10 V to +10'V. Incase one or both analogue inputs are not used, the associated input on the Centronics interface may be set to func- tion as a digital input, This is achieved by means of a jumper (C-D; E-F Each digital input consists of a darling: ton transistor (Ti-Ts), a collector resistor and tivo base resistors. The inputs switch at about +2.5 V, which makes them suit- able for driving both TTL and CMOS logic. The maximum input voltage is about +30 V The digital outputs, Tr-Ts, are of the ‘open-collector type to allow direct con- rection to small loads, such as LEDS or relays, Note, however, that any relay coil must shunted by a diode to prevent the transistor being damaged by inductive voltage surges. The maximum. voltage that can be switched by the output transis tors is 30 V. Two three-state buffers, IC and IC, and a LED array, allow the operations of the ADC and the DAC ta be checked vis: tually. Depending on the connection made by jumper G-H, the LEDs indicate either the data applied to the DAC, or the state of the digital inputs on the Centronics port. Since the monitor circuit falls short bf one line to provide an indication of SEL (select), this Is taken aver by two inver- ters, [CiICy, and LED Dis, The function of each LED is shown in Table 1 Construction ‘The converter is best constructed on the double-sided, theough-plated printed. cuit board of which the component mounting plan is shown in Fig. 3 (the track lay-outs are not given because this through-plated board is virtually im- possible to make without special equip- ment) The construction itself is precision work, but none the Jess entirely straight forward with reference to the component overlay and the parts-list. As already stated, the choice between the REF-02 and the TL31? is up to you: simply fitand omit the relevant components as indicated in the parts list. Connector Ki is a standard 3e-way Centronics socket with angled pins for PCB mounting. This type of con. nector is often used on matrix printers Control software and setting-up A 360-KByte MS-DOS formatted 54-inch diskette is available for this project. The programs on the disk are helpful for ad- justing and testing the converter board. A Turby-Pascal source file is provided that contains the basic routines for the [/O ‘operations with the A-D and the D-A con- verter. This program uses a set of default 1/O register addresses, which may have to be changed depending on the computer used, It should be noted that the logic levels in the status-, data-and control-reg- isters must correspond ta those of the in- puts and outputs. This point is made because the levels of the active-low lines are inverted either by hardware or soft- wate, again, depending on the type of computer. Obviously, the LED indicator array comes in handy here. Adjustment ‘The first routine given in the form of a flow-chart is the adjustment procedure (see Fig. #), All indicated voltages are meastized with respect to the analogue {ground potential. The relevant connection for the DVM may be found between the two analogue inputs and the analogue output FLERTOR ELECTRONICS: MAY 1990CENTRONIC 2O-QIOH +O-OFO-O% ynected to the circuit via Centronics socket Ki. The compute Fig. 2 Circuit diagram of the A-DID-A card. CTRONICS MAY TOR ELECOMPUTERS AND MICROPROCESSORS. eo0c0e000c00000000 © J ceecccce000000002 2 (290000000) “® oe ry ee nie nae? 2 AY “ore x2 csojjoo oojfoce Sik 21}00 coifo colo cay cr ig. 3. Component mounting plan of the doubl ihrough-plated PCB. First, adjust either Ps (relerence: is sufficient in Turbo Paseal- Sinee Un TL317} or Py (reference: REF-02) until a 3 V, the relation between the value of the relerence of 5100 V is obtained. Next, can datawond and the resultant analogue out cel the off-set voltage of ICs by adjusting Pa, Set the current that flows into the Vis input by adjusting P. Finally, adjust Pato Us = 3 (eata-128/128 (V} cancel the off-set voltage of IC put voltage is expressed by Writing data to the DAC is simplesin Data ‘(therefore produces -§ V; data nearly all cases, this involves loading one '80)/' 0 V; and data "FFy’ +4.961 V-Since regisier with the desired value, One state- the circuit is capable of producing 0 V, the ment, highest output voltage remains 39 mV below +5 V. A slightly different settin up procedure allows you to reach +5 V. PORT [DATAREGI = DATA: COMPONENTS LIS Resistors: 2 100k Rue 4 10k 8-way SIL Fe 4 4k0 1% Re 5. 10K 1% Frees Raw, Fs 2 20K1% Ree + tke Fe 1 4700 1% Fo 1 2200 1% Fo" 9 10% Ri-Ric 12 Bak Ra =F 6 ae oe — Ran Fuse 9 4700 Foe = Ri 4. 2kmultituen preset Pr 3) 10k multitum preset PsPa 410k mutitum preset arm +) 5000 mustitum preset Ps Capacitors: 5 1000 cut Gace 2 10FI6v Cs:Ge 1 120F cw ‘Semiconductors: 2 1NA001 Die 4 INSEE Da—De 9 LED Smm red Dr-Dis 1 LEDammgesn Ove 5 80317 4 BCsATC 1 TaHCTSS1 4 TAHETSd0 1 TaHeT4 + Tape + 7908 1 7a08 1 Tus17 1 RERO2 2 Far + Lissa 1 PM7524 (PMI) Kew 9° AD7523 (Analog Devices) Miscellaneous: 4 3@-way Centronics socket Kt jor PCA mounting; angled pins 1 26yayheader:angied Ke ‘ins; with eect handles 1. printad-circut board 1+ control program on disk 900097 1424 not ited when IGeis used notfited when Iris used but this in turn makes it impossible to achieve 0V, which can not be approxt mated ata difference better than 39 mV or VALSBat data = $0). In practice, i is easier to state a voltage and calculate the corre sponding DAC data from da 541-3 < Uys 4.901) 28 (U whore ‘data’ is counded aff to givea whole number. Both the checking of LI, and the required conversion computation to pro- vice the necessary bit combination for the ELEK CTRONICS MAY 1990CENTRONICS \-DD-4 CONVERTER Fig. 4. Flow diagram of the adjustment procedure. DAC may he included in the routine that controls the voltage setting, A-D conversion As already noted, this is effected on the basis of successive approximation. The unknown analogue voltage is approxi- mated by comparing it to analogue volt- ages generated by the DAC, whose resolution causes the longest (worst-case) approximation to require the maximum, number of steps, 2° ur 256. The use of different approach may reduce this num- ber to 8, oF the width (in bits) of the data input of the DAC. Figure 5 shows the flow-chart of the D-A routine. This proce- dure is invoked with two variables: the input channel and the attenuation (for which variable K must be greater than 1) BLEKTOK FLECTRONICS MAY 199) Fig. 6. Direcioty of the PO-MSDOS pro- ‘gram disk, number 1421, tor the project. Variable ‘data’ is used for intermediate data storage, starting with value 0. Next a for-nex! loop is enivred. This is passed ‘ight times, during whick the A-D conv sion is effected, A program within the loop checks for gach bit whether thismust become (or 1. ‘The most signiticant bit is treated first with an intermediate value of Dall bits are 0). The sum of these two is, obtained with the aid of an OK function and is subsequently written to the DAC. When the relevant comparator indicates that the voltage is too high, the corre- sponding bit in the intermediate value ‘must remain at 0, When the voltage is 100 Jow, the same bit must become 1. Alleight bits are treated in this manner by shifting the 1 to the left (GHL). After the eight steps, the input voltage may be calculated ‘on the basis of the intermediate value Short wait times are inserted between the write operation to the DAC and the read operation to the comparator. This is done to allow for the response time of the ICs, This wait time is so short, however, as to make the use of standard time func- tions in the PC impossible, since their minimum delay of about 1 msis much too long. Hence, a for-next loop is used. An ‘obvious problem caused by this approach, dependency on the clock speed of the computer, may have to be resolved em- pirically with different loop repetitions. In many cases, a single repetition is sult Fig. 5. Flow diagram of the successive approximation routine cient to establish the required wait time In that case, the loop may be replaced by ‘one oF more useful stalements of your own. A statement like ‘repeat until true’ does effectively nothing but last a number ot clock eycles and does not require a pre- viously declared ausiliary variable ‘Test program The diskette supplied for this project con: tains the basic routines in a Turbo-Pascal unit, both in the form of compiled code and source text. Also on the disk is an auxiliary program for testing and adjust: ing the card, This progeam, CENTREME-ENE, searches for a file called CONFIG.DAT, which contains five numbers that indicate the printer port number, the attenuation on input channel 1, the attenuation on input channel 2, the reference voltage, and the number of iterations in the wait oop, in that order, This configuration file may be edited to individual requirement with the aid of any ASCH compatible word processor, like EDLIN or the ane in SideKick or PCTools. The numbers are separated cither by a comma or a space. The filealso contains a few linesto explain the meaning of the numbers. These lines are comment and have nothing to do with the actual operation of the test program, which, incidentally, may be run without the A-D/D-A card connected to the com- puter. This is particularly useful to become acquainted with its structure and commands. .SCIENCE & TECHNOLOGY IMAGE SEGMENTATION MS. Kishore The aim of this article is to give a basic idea of Image Segmentation techniques and how they are applied to a given image. Several techniques are described: it is shown that the use of each of these depends on the specific requirement. Segmentation 15 a technique for splitting aan image into regions that hopefully repre: sent the surfaces in the real world where the image originated. Its purpose is to ere ate, by algorithms, « symbolic representa tion of the scene rather than the pixel ‘we normally look at. If, for instance, we ‘want fo express a scene consisting of a cir cle of radius r. centre (a,b), intersecting: & square of side s, centred at (4, 9). We Want the segmentation process 10 dispense with the usual pixel-based information and give tus the parameters of the scene in a more concise and meaningful way S tant elements in automated image because jt enables objects or other aspects of interest to be extracted from an image for subsequent processing, such as de scription and recognition, Segmentation algorithms are generally based on one of the two basie properties of ‘grey-level values: ciseonrinuity and sini larity. An image that is based on abrupt changes in grey level 1s classed in the first. The principal areas of interest in this, category are the detection of isolated Points and the detection of lines and edges ‘The main approaches in the second cat- egory are based on thresholding. region growing and splitting and merzing. ‘The concept of segmenting an image based on discontinuity or similarity of the grey-level values of its pixels is applicable ‘o both static and dynamic (time-vartable) mages. In the latter case, however, mation can offen be used as @ powerful cue to in prove the performance of a segmentation algorithm, mentation is one of the most impor: Point detection ‘The problem of detecting and then sew. menting isolated points in an image ap. plies in noise removal and partic sis. The basie mask used for detecting iso: Fated points in an image is -1] -1] -1 -1] 8} -1 -1 | -1} -1 Aveach mask location, we compute the Vector product IDB BST B HD, In level, the result Of this operation would have heen zero. Since in the example the image is centred at an isolated point (8), where the inten- sity 1s greater than at the other locations, the result is greater than zero, In practice, where one is interested conly in strong responses, we say that an isolated point, whose intensity is signifi- cantly different from the background, has been detected if the vector product is, reater than some non-negative threshold. Line detection The neat level of complesity involves the detection of lines in an image. Consider the mask “1 4 Bil) Bs) +8 a} a) horizontal lines: with constant-tevet pack ‘ground, the maximum response would ob- tain when the line passed through the mid dle row of the mash ‘A similar experiment would reveal that this mask would respond best to lines at +45%: this mask mT 2) -1 =} 2) -1| 2) —-1 to vertical fines: and this mask 2 -1 -{ 2/4 -|-1| 2 to lines at 45° The direction of the lines may also be established by noting that the preferred di reetion of each mask is weighted by a larger coefficient (i.e., 2) than other poss: bile directions. ELERKTOR ELECTRONICS MAY 1990Edge detection Although point and line detection certainly are elements of any discussion on segmen- tation, edge detection is by far the most commonly used approach for detecting meaningful discontinuities in grey level. The reason for this is that isolated points il thin fines are not frequent occurrences in most applications of practical interest In this approach, we define an edge on the boundary between two regions with relatively distinct grey-level properties. It is assumed that the two regions are sutfi- ciently homogeneous for the transition fiom one te the other to be determined on the basis of grey-level discontinuities alone. When this assumption is not val line oF point techniques are generally more suitable than edge detectio Most of the edge detection techniques involve the computation of a local deriva- tive operator. The first derivative of an edge is zero in all regions of constant grey level and assumes a constant value during a grey level transition, ‘The second derivative is zero an all lo cations, except at the onset and termina tion of grey-level transition, Tis evident that the mage first derivative ean be used to detect the presence of an edge. while the sign of the second derivative may be used to deter- ‘mine whether an edge pixel lies at the dark (background) or light (object) side of the edge. The sign of the second derivative is positive, for instance. for pixels lying at the dark side of both the leading and trail: ing edges of the abject, while the sign is negative for pixels at the light side of these edges. Similar comments apply 10 the case of dark object on a light background, The direction of the gradient vector is also important. ude of the Gradient operation As indicated, the gradient of an image fix, y) at location (x. y) is defined as the two-dimensional vector Gliis.y)] = [Gx Gy] tis well-known, that the vector G points in the direction of maximum rate of change of f at location Lx, »). For edge detection, we are inter ested in the magnitude of this veetor, ger erally referred to simply as the gradient and denoted by Gfx. y)], where ELERTOR ELECTRONICS MAY 1990 Gif, 9] = 1Gx + Gy] ‘This quantity is equal to the maximum rate of increase fly, y) per unit distance in the direction of G. The direction of the ‘gradient vector is also an important quan= tily. IF (x,y) represents the direction angle of G at location (vy). it Follows trom vee- {or analysis that Gav.) = tan (GIG), where the angle is measured with respect to the axis, Consider the sub:image area xd | x2] x8 x4 | xB] x6 x7 | x8] x9 here 5 represents the grey level at loca tion (4, 9) and the other mask locations represent the grey levels of the neighbours of fix.) We define the component of the ‘gradient yeetor in the « direction ay Gx = (07 + 2s8 429) — (0 +2 453), and in the y direction as Gy = 03 + 2x6 449) 0h + Dd EAT), The use of a 3.3 area in the computation of the gradient has the advantage of in: creased smoothing over 2x2 operators. tending (0 make the derivative operations less sensitive to noise. ‘Weighting the pixels closest to the cen- tre by 2 also produces additional smooth- ing. It is possible to base gradient compu ations over larger neighbourhoods (Kirsch), but 33 neighbourhoods are by far the most popular because of the advan tage in computational speed and modest hhandware requirements, It follows from the discussion in the previous (wo sections that Gar ean be com- puted by using the mask at | 2 G 0 0 0 1} 2] 4 SCIENCE & TECHNO cy-|-2] o] 2 These 1wo masks are commonly 1 Ferred to as the SOBEL operators The responses of these operators at any point (4, 9) are combined to obtain the era dient at that point. Convolving these ‘masks with an image f(x, v) yields the gra- dient at all points in the image: the result is often referred (0 aS a GRADIENT image. Another useful approximation, called the ROBERTS gradient. makes use of th cross differences given by Gifix, yi] = [foe yy five Loyd + + [tiv + Ly) fey pT The LAPLACIAN is a second-order derivative operator. which ean be imple ‘mented by convolving the mask 0 1 Oo 1] -4) 4 0 1 0 Although the LAPLACIAN responds te- lransitions in intensity. itis seldom used by itself for edge detection. The reason for this is that. being a second derivative oper- ator, the LAPLACIAN is typically sensitive 10 noise. It is, therefore, usually relegated 10 the secondary role of serving as a detect for establishing whether a given pixel is a the dark or the light side of an edge. ‘The vector formation for the detection Of points, lines and edges has the impor tant advantage that it ean be used 10 detect combinations of these features. The tech: nique was developed by Frei und Chen. The nine associated masks are shown on the next page: the first fourare suitable for detecting edges, the second set of four rep resemts templates suitable for line detec tion and the last mask is proportional (0 the average of the pixels in the region at which the mask is located in an image. Gradient image threshold If we take the threshold of @ gradient image at a moderately grey level, we find both object and background below thresh ‘old und most edge points above threshold,Be Kirsch has developed a method that makes use of this phenomenon. In this. the gradient image is first given a moderately low grey level threshold to identity the ob- ject and the background, which are sepa. rated by bounds of edge points. Then, the threshold is increased gradually. which causes both the object and the background {0 grow. When they touch, they are not al- owed to merge, but the points of contact, ‘which define the boundary, are noted. This method is computationally expensive, but it tends 10 10 produce maximum gradient boundaries while avoiding many of the problems of gradient tracking bugs. For multiple object images, the segmentation is correct if. and only if, itis earried out ‘correctly by the initial thresholding step. The edge operater developed by Kirsch also detects the presence of edges. It func~ tions as follows. Each 3x3 neighbourhood is convolved with eight kernels. The maxi- ‘mum value over each of the eight orienta: tions is taken as the output value SEGMENTATION Segmentation by thresholding Thresholding is a particularly useful tech- rhique for scenes containing solid objects resting on a contrasting background. It is, computationally simple and never fails 10 define disjoint regions with closed con- nected boundaries. When using this tech- nique for image segmentation, one assigns all pixels at or above the threshold: all those with a grey evel below the threshold fall outside the object. The boundary is then a set of interior points, each of which has at least one neighour outside the ob- ject Thresholding works well if the grey level of the object is uniform and the ob- Ject rests on a background of a different. but also uniform, grey level If the object differs from its 4] v2] 4 +} of a o| -1| v2 Oo} o}| o v2] 0|-~2 A, |} O91) 4 4 [2] -1 1] of we] 1] 0 v2|-1] 0 o}| i] 0 -1| 0} 4 1} of} 4 1} o| 4 o}| o] o o{ 1/72 o]| 1] 0 1] Oo} 4 1 1 -2} 1| 2 ff. a peca 2] 4| 2 +] 4] 4 a] ay a 1 1 -2| 1] -2 1] 1] 1 Masks used inthe Fret-Chen technique, rey level was selected, Adaptive threshold: in many cases the background grey level is not constant and the object contrast varies with the image. In such cases, a threshold that works well in one area might work poorly in other areas of the image. It is then convenient to use a threshold grey level that is a slowly varying function of the position in the image. Optimal threshold: unless the object in the image has very steep sides, the exact value of the threshold grey level can have 4 considerable effect on the bound sition and overall size of the extracted ob- ject. For this reason, we need an optimal. ‘or at least consistent, method for establish ing the threshold. An image containing an ‘object on a contrasting background has = bimodel grey level histogram. The two peaks correspond to the relatively larger umber of points inside and outside the object, The dip between the peaks corre: sponds to the relatively few points around the edge of the object and is commonly used to establish the threshold grey level po background by some property other than grey level (such as texture), one can first use an +5 +5) +5 +5 | +5 +5 operation that converts that property to grey level +5 +5 +5 In the simplest implemen: tation of boundary location by thresholding, the value of the 45 +5 threshold grey level is held constant throughout the image. If the background grey +5 level is reasonably constant throughout the image, and the object has a roughly equal & 45 contrast against the back- ground, a fixed global thresh~ old will usually work well, +5 +5 +5 provided a correct threshold Results of convolving a 3x9 neighbourhood with eight Kernels. ELERTOR ELECTRONICS MAY 1990,Region growing by pixel aggregation Region roups pixels or sub-regions into larger regions In its simplest form, pixel start with a set of “Seed” points and from these grow regions by appending to each sub-point those neighbouring pixels that have similar properties, such as grey level texture, and colour. For instance, in applications of infra- red imaging, hot targets appear brighter } than the background. Choosing the bright- est pixels is then natural starting point for a tegion-growing algorithm, The selection of similar criteria de pends not only on the problem under con- sideration, but also on the type of image data available. For example, the analysis of satellite imagery is dependent mainly oon the use of colour. The analysis would be much more difficult if only mono chrome images were available All boundaries between adjacent re- gions are examined, A measure of the boundary strength is computed from the differences of the averaged properties of the adjacent reg siven boundary is strong if the properties differ significantly on either side of the boundary and weak if owing is a process t they do not joundaries are allowed to stand, boundaries are dissolved and the adjacent regions merged. The process is repeated by first computing the object membership properties of the enlarged re- gion again and then dissolving the weak boundaries. The process is then continued tuntil a point is reached where the bound- aries are weak enougt to be dissolved, Region splitting and merging ‘The region growing process starts from a set of *seed" points. An alternative is 10 sub-divide an image inititally into a set of arbitrary. non-joining regions and then merge or split the region in an attempt to satisfy the requirements discussed earlier Given a digital image containing sey- cral objects. the pattern process consists of three major phases. The first of these is ‘object isolation, in which each object must be found and its image isolated from the rest of the scene. The second is called fea ture extraction, The features are formed by ‘a set of measurable properties, The extrac tion phase measures these properties from which it produces a set of measu called the feature vector, This drastically reduced amount of information represents SCIENCE & TECH cry oLoGy all the knowledge on which the subse- quent classification must be based. The third phase is object classification, which is merely a decision as to Which class the ‘object belongs ‘When a human observer views a scene, the neurological process that takes place in the retina and the optic cortex essentially segments the scene for him. This is done so effectively that he sees not a complex scene, but rather something he thinks of as a collection of objects. With digital pro- cessing, however, we must isolate the ob: jee fe by breaking up that ‘image into sets of pixels, each of which is the image of one object . References: Digital Image Processing by Gonzalez and Wintz: Addison Wesley 1977. Digital Picture Processing by Rosenfield and Kak. Academic Press 1982. Digital Image Processing. by Castleman, Prentice Hall 1979 Digital Image Processing by W.K. Pratt, John Wiley & Sons 1978, Imeractive Digital Image Processing Soft- ware by M.S. Kishore (Dissertation 10 University of Poona, 1986). (A) Original 64564 1age; (8) after applying the Sobel operator Gx; (C) attr applying the Sobel operator Gy: (D) after applying the Sobel operator G: (E after applying Roberts gradient operator: (F) ater applying horizontal lin detection mask; (G) after applying vertical line detection mask; (H) after applying Laplacian operator; and () after applying edge detection. ELEKTOR ELECTRONICS MAY 1990, ea eeeBUDGET SWEEP/FUNCTION WW - \ be GENERATOR Y\ WY T. Wigmore This month we add yet another item to our series of budget test instruments. The signal generator described has a built-in sweep function which is ideal for audio measurements. Based on the well-known XR2206 function generator IC with very few external components, the instrument offers a hard-to-beat Itis not wise to disregard the XR2206 from Exar when designing an all-round func tion generator. The device is versatile like fno other, and guarantees a fairly simple cireuit for the given application. Further ‘more, its cost makes any attempt at desig ning an equivalent circuit based on discrete components a waste of time while its output signal distortion figures are not spectacular, but none the less low ‘compared to those of a competitive chip like the 8038. For use as a basic function generator the XR2206 requires only a handful of passive parts. The frequency adjustment ‘nd the sweep function are simple to im plement by the addition of one dual opamp and three transistors. The output amplifier of the instrument also follows the general line of comprising of as few components as possible: only one p opamp is required The generator The XR2206 forms the heart ofthe circuit se Fig. 1), With the external components Configuration used here, the IC supplies a Sine-waveanda triangular waveat output price/performance ratio. pin 2. The d.c. operating point sset tohalf the supply voltage (6 V at pin 3) by poten: tial divider Ris Rio. The resistance at the potential divider junction, 16.5 k@, and the voltage at the AM input, pin 1, (0/V) determine the amplitude of the output sig: nal The wavelorm selection is effected by one contact af Ss. [n the position shown, resistor Rit is connected to pin M4 of he XR2206. The current flow through Ri en ables the IC to convert the triangular sig- pal into asi The value of Ri determines to what estent the inflection points of the triangle are rounded ta give 2 sine-wave. For the sake of simplicity, a fixed resistor Instead of the expected (multiturn) preset is used to set this cur rent When the contact of is opened, pin 2 supplies a triangular signal whose peak amplitude is twice that of the sine-wave The rectangular wave is supplied via pin 11, This open-collector output of the XR2206 js pulled to ground at the gener ator pulse rate by an n-p-n transistor. Voltage divider Ris Riv-Rs) at pin 11 sets the amplitude of the rectangular wave The maximum and minimum voltage le MAIN FEATURES Frequency ranges: 3 (10 Hz ~20 kHz) ‘or § (10 Hz 200 kHz) ‘Sweep frequency: 0.1 Hz=100 He ‘Sweep range: 0-120 Sweep output: 5 Vest sawtooth; Za 1ka Waveforms: sine-wave, triangle, ‘rectangle Distortion (sine-wave): 0.5% yp. (in AF range) [AC output: all waveforms; Zo = 50.1, ‘shor-ciroult resistant ‘Output amplitude (A. = 30.2): 1.1 Von ~1 Vpg (sine-wave) (0.1 mVpp ~ 2.5 Voo (triangle) 10.1 moo ~ 1.5 Vpg (rectangle) Output amplitude (A = 600 2): 0.1 mVip to 1.8 Vip (sinewave) 0.1 mop 4.5 Vp (triangle) (04 mVmp~3 Vou (rectangle) Currant consumption:approx. 100mA atv ELEKTOR ELECTRONICS MAY 1990BUDGET SWREPFUNCTION GENERATOR Fig. 1. vols are 9,1 V and 3.8 V respectively. This swing is close to the optimum drive mai gin of the power opamp that follows the XR2208. The second contact of Si selects either of the two IC outputs and passes the rele vant waveform to the output amplifier. Frequency control The frequency of the signal supplied by the ¥R2206 is determined by two factors: the capacitance between pin 3 and pin 6, and the current drawn from pin 7 The capacitance is determined by the three capacitors selected by the frequency range switch, Ss. A fourth range (up to 200 kHz) may be added by providing an extra switch position and a capacitor of 2.2 nF (see the section on construction fur- ther on) Pin 7 of the XR2206 supplies a tem- perature-compensated reference voltage 6f3V, which is also available at pin 10, where it is decoupled by Cs. The voltage at pin 3is loaded by a resistor, Rus,and the output of opamp ICis. Hence, the output voltage of the opamp determines the cur- rent through Ris and with it the signal frequency, ELERTOR ELECTRONICS MAY 1990 f= du 3C where fis is in ampares. Factor C is the capacitance (in farads) between pins 3 and 6. Frequency and frequency sweep ad jnstment are effected manually by poten- tometer Ps at the -input of ICim. When 5: fs sot to the ‘normal’ position, Py and Re form a potential divicler that limits the voltage at the wiper te a value between DV to V. Resistors Ri-Ryy set the ampli fication of the inverting upamp to a value that results in output voltages of virtually OV and 3 V sith Py set to maximum and minimum (wiper to ground) respective The dc. operating point —and with it the start of the frequency range — is deter- mined by PeRis and Ris, which ensisee that a part of the -V reference voltage is, applied to the non-inverting input of IC. Sweep function When the generator trequency is set ma. ually, a fixed resistor, Re, provides the direct vollage to potentiometer Pi, When S> is switched to the other position, how ever, Ps is supplied with the output volt ‘Circuit diagram of the sweep/lunction generator. The Heart of the eiroull is formed by 1G2, an XR2206 trom Exar. age ota ramp generator. In this mod potentiometer sets the swept frequency range rather than the frequency itself. In other words, it determines to what extent in H2/V) the ramp generator can change the set generator Irequeney The ramp generator is formed by opamp IC), and integrator C2. The inte gration time is set by the voltage at the wiper af Pi: the higher the valtage, the faster the capacitor is charged, and the faster the sawtooth voltage rises. Poten- tometer Pr allows the sweep time to beset to a value between 10 ms and 10s. The ‘iaximum time is calibrated by preset Ps which also serves as an off-set compensa- tion for IC The rise of the sawtooth voltage at the Integrator output is ended via Ts and T) The emitter of Te is held at a reference potential provided by zener diode Dy. The Iransistorconduets, and T) and Tiare kept tif, as long as its base voltage is below the reference. Assoonas the saivtooth reaches a level of about 0.5 V below the reference voltage, T2 is briefly turned off, so that its collector voltage is pulled to about DV via Ri, Asa result, To conducts and resets the integrator by making the inverting input of ICs positive with respect to the non-in-B TEST AND MEASUREMENT FLEKTOR ELECTRONICS MAY 1990Cee 2 Pyle + 19a Re 2 2200 Rai La 2K Re 3 10h RsRafao 1 1Mo Be 2 10 Rape 1 Bk a 2 470k Brefis 2 820k Rove 1 oe Fa 2 3% Fisie 1 2% Rie + 1908 Rev 2 1000, Fania ake Re 1 100k1in. potentiometer Pr 1 250R preset H Pe 1. 50klin, potentiometer Ps 4) 100k preset H Ps 1. 10ktin. potentiometer Ps ‘Capacitors: + ang o 1700 Ge 4100" Cabecisor 1 8p ce 4 1n016V.axial Gs 1 22 MKT o 1 Zz Ge 2 2200 ran 1 a2 Gro 1 2r2 (optional) Ce 1 W0aF 18Vaxial =O 1 400uF i6V axl = Ge 4) ATOOMF 16Vaxial Crs 1) 1000uF 25Vaxal Cus ‘Semiconductors: 7 5¥e <00mW or ener diode + qNéo01 oe + LED bs 1 BS250 % 2 BOSs7B Tats 4 CAS2ADE Ics 1 xR2206 Ia 1 Lies ics 4 yee Ge Miscellaneous: 1 miniawre SPST switen” Si 4 miniature SPOT swich © Se 2 Boole 3-way snitch S18 (igniter MTAZO6PA or CAK'7211) ($2 may be replaced by a miniature “pole way rotary wich) 2 BNC socket Kuk 2 0-220 style heatsink + printed-creut board ‘s00040 1 front panel fi -900040-F Construction The single-sided printed-circuit board on which the generator is constructed. is shown in Fig. 2. Population of the PCB is straightforward with the possible excep- tion of the following points: 3 l2 3 2 a 4 Sf ols wy Ga 6 5| 40 60 5| D | 5| al dO 900040 -4 Fig. 3. ‘+ ‘ThespindlesofpotentiometersP), Peand Pyare inserted from the track side of the PCB toenablethenuts on theshafts tobe locked at the component side. Use short wires to connect the potentiometer ter- minals tothe relevant copper tracks + Asshown in Fig, 3, Csand [Ci are fitted atthe trackside ofthe beard. Do observe their correct orientation and the electri- cal insulation of the heat-sinks. «+ Switches $), Si and Si, and the BNC soc kets are mounted on the front panel. Their pasitions correspond to those pro- vided on the overlay printed on the ready-made circuit board. The connec tions are made in shart lengths of light- dluty insulated wire It is not strictly necessary to use IC soc kets, although thesmall additional invest ment may prove worth while ifa faulty IC is suspected. Since the instrument has its ‘own single-phase rectifier, smoothing ca pacitor and 12-V voltage regulator, it may be powered from an unregulated AC or DC supply with an output oF 15 V to 18 V, @ transformer is used, observe the necessary safely precautions as regards insulation of the mains voltage and the fuse rating, Setting up It is recommended to adjust the com. pleted printed-circuit board before it ts Fitted into the enclosure. This means that the switches and the output sockets have to be connected provisionally. Apply power and allow a few minutes forthe circuit to warm up. Set 82 to NOR- MAL, and Ps to a frequency roughly at the centre of a range, ex. 100 Hz. Connect a frequency meter (o the signal output and adjust Pr until the measured frequency ‘equals that Set on the scale. Iryou donot haveaccess toa frequency meter to perform this adjustment, use the beat frequency method instead. Feed the 100 Hz signal obtained with the aid of a small mains transformer, a bridge rectifier ful series network of a 100-82 resistor ‘Connections made In the 3-position switches from C&K. and a 100-pF capacitor, to a loudspeaker Drive another loudspeaker with the gen= erator output signal. Listen to the two i nals and adjust Pi for zero frequency difference. This method gives quite accur ate results (For use with a bO-H1z mains, set the generator to 120 Hz), The adjustment of the sweep function is carried out at the greatest sweep time 10s. Turn Py fully counter-clockwise and connect an analogue voltmeter or a LED to the sweep output of the instrument ‘Adjust P> until a time period of 10 s is obtained Tips and options In the basic arrangement, the waveform and frequency range selection are effected with 3-position miniature switches trom C&K. The switching configurations are shown in Fig. 3. In the case of $5, the use ‘900080-5 Fig. 4 Allernative switch connection which enables the trequency range of the generator to be extended to about 200 kHzTEST AND MEASUREMENT Fig. 5. of a switch that has only three positions limits the frequency range of the instr ment to about 20 kHz, A fourth range which may be desirable in a number of cases, may he added by replacing the toggle switeh with a small, four-position, rolary switch, which is wired as shown in Fig. 4, The numbers 1 to 6 on the overlay mark the connections of the terminals of Ss. To create a 200 kHz range, solder a ditional 22 AF capacitor, Cin, to the centre terminals, numbers 2 and 5, and der a wire between terminals 2 and 4 Next, connect the contacts (1, 2and 3) and the pole of the rotary switch to the PC terminals 1, 3, 6and 5 As already noted, the value of R determines the shape of the sine-wave. At LOW-BUDGET TEST EQUIPMENT! This is the fith instalment in a series of articles describing test equipment no serious electronics enthusiast or dasign engineer can do without. Ail | Instruments are housed in an attrac- tive metal cabinet type LC-850 from Telet, which comes with with protec- tive strips at the sides. The switch areas on the front panels are grey. light blue or dark blue with white let: tering, and their size Is geared to the front pane! of the LC-850 enclosure. Shown in the picture are the instru ments described so far in this series The power supply stown in front will be next month's subject. The pile of four instruments behind the sweepifunction generator con- sists of (top to bottom): + LEIHF signal tracer (January 1990) + Qmeter (May 1990) + RF inductance meter (November 1989) + ACmillvoltmeter (February 1990) ‘The pointer knobs used on the instru- ments are made by applying a small arrow or triangle (available as transfer symbols) on to the collet and protect- ing it with plastic spray. relatively high generator frequencies, it may be useful to replace the resistor by a 500-42 preset to enable the distortion to be minimized. Br tests, the XR- om a number of practical 106 supplies a fairly clean ‘up to about 100 kHz, To: wally sinusoidal sig wards 200 kHz, the sinewave grad changes into a triangular waveform. The L165 is capable of providing con siderably more output power than it is allowed to by the 50-2 output. If it is fesired to use the generator forswept-h dlesired to use the g pt-fre {quency measurements on loudspeakers 0 drive units, 9 losv-impedance output may be provided on the instrument by fitting two binding poets on: the-rear panel. The jgnal outlet is connected direct to the ne gative terminal of Cis to negate ‘Completed PCB-front panel assembly seen from the PCB track side (left) and from the side (right). esistors, Note requires a rather larger power supply. In that or lext, it is recommended to use of the two 100-4 however, that this extension transformer capable of supplying at least srrent,a bridge rectifier (000 FE single-phase 1A ofsecondary (x1N4001) and an addition apacitor. ‘The rectitier on the board, D2, is replaced by a wite link. The I-A power supply enables the function driving power for kers. The use of a bridg mnerator to provide ample Toudspea ifier instead of the single-phase rectifier allows a mains transformer with a secondary volt age of 12 V to be used instead of a 15-V ype : TOR ELECTRONICS MAY 1990TRANSISTOR CHARACTERISTIC PLOTTING S. Aaltonen The circuit described here makes use of a computer to plot the so-called output characteristic and determine the small-signal current gain, hie, of an n-p-n transistor. These two transistor parameters are of great importance for classifying an unmarked transistor, for a reliable good/faulty test, and for selecting matched transistors from an available lot. Although the program that controls the circuit is written for the Atari ST series of home computers, the use of the Centronics port should enable owners of other micros to adapt their own version fairly easily. The use of @ computer and a printer, i stead of the more usual oscilloscope, to measureand record transistor parameters js subject to one imporiant proviso: the transistor under test must be located be tween a digital-to-analogue convert (DAC) and an analogue-to-digital conve er(ADC). The circuit presented here has, therefore, a digital inputas well asa digi tal output, both of which are connected the Centronics (parallel) printer port 10 convey the necessary data and control le- vels to and from the computer. Transistor parameters Since the basics of transistor characteristic plotting have been covered relatively re- cently im Ref. 1, only a recap is given here Figure 1 shows an ideal transistor in the standard four-polle test circuit in Which voltages are applied to the base: cemitler junction and the collector-emitter junetion, The first important transistor par ameter that may be obtained from th basic test circuit is the so-called output characteristic, which is a curve that de- scribes the relation between the collector emitter voltage, Ucy, and the collector current, Ic, with the base current, [y, a8 a parameter. Ideally, such curves are straight lines since the collector current is determined hy the base current only, and nat by the collectoremitter voltage. In practice, however, the so-called early et fect causes the Ic-vs-Licy characteristic t0 become a curve rather thana straight line test eireult Fig. 1. Transistor in a four-pol at relatively low values of particularly Ucs The second important characteristic is the small-signal current gain, bye. This is defined as the ratio of change in collector current, fc, to the change i base current, 8ly, that produces i, when the collector: emitter voltage is kept constant: h Nc Bly = fe/ly When Lce is constant Most transistor manufacturers provide this parameter at two or three values of Ucr-and ty The present circuit plots the output characteristic of n-p-n transistors for eight values of Jy. and in addition automatically calculates a statistically derived hyy value With these two parameters on the sereen and on paper (hard copy trom the printer) you are in a position to select matching transistors for critical applications, or find a substitute for an unknown transistor Circuit description Asalready stated, the transistor undertest (LUT) is located between a DAC (1C2} and an ADC (ICa) —see Fig. 2 All control and processing of measures! values is carried put by the computer, ‘The circuit Uses two supply voltages! 5 V for the DAC, the ADC and counter Cy, and 15 V for the transistor test circuit and the associated voltage amplifiers. The FLERTOR ELECTRONICS MAY 1990higher supply level of 15°V is required to provide the TUT with a maximum collec~ toremitter voltage of about 9 V. The meastirement is cyclicand control: led by the computer, First, the base cur rent of the TUT is set at a certain value. Next, the collector-emitter voltage is tmised gradually from 0 V to about 9 V, and the resulting collector current measured. This process is repeated with the next higher value of the base current. The step size is 25 1A, and there are eight steps starting at fy =0 1A. The test cycle is complete at fy = 175 WA. The control program provides a stream of elock pulses on the DI (data-1) line of the Centronics port. The clock pulses are counted by IC, a Type 7HHCTAMO, The counter yaiues at the Q0-Q7 outputs are converted to an equivalent analogue volt- age between () V (value: 0) and 25 V Walue: 255) by DAC ICs. The QS, QS and ‘Qi0 outputs of ICr control the base cur rest of the TUT in 8 steps, The required current step sizeof 25 1 is obtained with the aid of resistors R-R. Note that the Jalue oF equivalent value of each resist ance at the three counter outputs is derived from 180 kO, since this value re sulls ina current flow pf about 25 uA at a logic high voltage of about +43 V at the respective counter outputs “The clock pulses provided by the cor puter eausette voltageat the outpatot IC {ibe increased from 0'V to the reference voltage of the 2425 (2.5 Vo in 255 steps, Initially, this happens with Q8, 99 and QW of the counter being lowe so that le — DKA. The analogue voltage is amplified by a factor of four by opamp ICs. The resulting Voltage range at the collector of the TUT is about 0'V to 9 V. This voltage range is divided by twa by Ris-Riv ta pre~ vent the maximum input vollage of ADC ICs being exceested The emitter current of the TUT causes avoltage drop across Ris. This voltage is amplified by a factor of 48 by opamp IC before itis applied to the AO input wi the ADC. Note that the emitter earrent rather FRANSISTOR CHARACTERIS cornne yg ra TEU PUTT TT TTT Ty acini sca=a040e Fig. 2 Circuit diagram of the compuler-controlied transistor curve tracer. than the collector cursent of the TUT is ‘measured. This ean be done without prob: lems, however, since in the four-pole test circuit the emitter curzentis thesum of the collector current and the base current. The latter is in the WA range and is, therefore, negligible with respect to the callector current, whieh is in the ma range Every time Uce of the TUT reaches its ‘maximum value of about 9 V, its reset to 0-V again, and the base current is in creased by 25 MA, to start a new curve. Fig. 3. Block diagram af the TLCTS41 and the pertinent pulse liming on which the control program flow is based (ilustration reproduced by kind courtesy of Texas Instruments) ELRKTOR ELECTRONICS MAY 19%TEST AND MEASUREMENT % a a '000000000000000000 = 999999000008000000 3 ° Fig. 4. Track layout and component mounting plan af the single-sided printed-circull board This eycle is repeated eight fimes until a high pulse on Centronics line D3 resets eounter IC Processing the analogue quantities The Type TLCIS41 (ICs) from Texas In- struments is a 10-bit, TI-channel_ anal: ‘ogue-to-digital converter with an internal analogue multiplexer and a serial data ‘output. In the present circuit, only two of the available 11 channels are used. One channel, AO, takes the Je parameter, the ‘other, A3, the Uy parameter. Figure 3 shows the block diagram of this interesting LinCMOSehip,along with the pertinent timing sequence. The computer selects the channel from which it requires the 1)-bit data. This se lection is accomplished by’ pulling CS of the TLCIS41 low via Centronics bit DO and applying the relevant channel code (0 for 3) serially: via Centronies bit DS, All channel selection, fim serial_data. output operations in. the TLCIS41 run under the control of SVS: CLOCK and (OCLOCK, tor which the re quired pulses are’ supplied by the computer via Centronics bines D2 and Da respectively. The 10-bit output data for processing by the computer is shifted out serially with the MSB first, The conversion err of the TLCI541 js +1 LSB, or 5 V/ 1024 4.8 mV ata maximum voltage of 5 V at the channel inputs. Hence, the maximum error of Ucr is about 10 mV, which is- ac ceptable in the present application, The ig, conversion, and COMPONENTS LIST Resistors: 10 1808 RuROR FaReRs 47K Fam 27k Fe-Rio 30 Fir 700 Ay 2106 RusiRs Capacitors: 4100 roy 220MF 25V cs ‘Semiconductors: 1 C5478 t 1 7aHCTAD40 Ics 4 ZNi25 ice 4} CAsQM0€ ies 1 TLO1S41 ice 3 INea8 D003 ‘Miscellaneou: 1 9G-wayPCB-mount Ki ‘Cenironics socket + PCB 1. contre! program ‘900058 ESs1491 computer reads the measured value by monitoring the state of the wUsy input line on its Centronics port Control program The control pro ram for the curve tracer + provideclock pulses tothe Lice-fe gener ators + arrange the timing sequence of the TLCISH + rend the measured values of Ucpand le that belong with a particular value of Is + calculate an average hye value + plot icas a function of Uce with Ip asa parameter + providea gr + allow the dumped tc shies screen raphs on the screen to be printer to obtain hard copy AIL this is arranged by a program written in C for the Atari ST senes of computers, This program, npr.prg, and the source file, npn, are available on disk. A. few examples of output characteristic plo are shown in Fig. 5 Construction and use Construction of the computer-controlled curve tracer is straightforward if the printed-cireuit board shown in Fig. 4 is used. Connector Ki is a standard 46-way Centronics sacket for PCB mounting. As shown on the photos type. plastic PCB spacers. this connector 1s mounted on two An alternative that does not require spacers is a similar con- nector with angled terminals. Both types of connector are often referred toas lue- rilbon’ and are commonly used on matris printers, ELERTOR ELECTRONICS MAY 1990‘TRANSISTOR CHARACTERISTIC PLOTTING 9] et wo term we ns n a} “ # 0 ry ry Te 108 * Fig. 5. A few examples of plots made by the program. Start the construction by fitting the seven wire links, followed by the IC so kets. Next, mount the resistors, the capa- Citors, diodes and the single transistor. A transistor test socket may be ised for inserting the TUT, but in many cases three light-duty flexible wires with small, plas- tic covered crocodile lips are perfectly all Feht The ICs are fitted last. Observe their orientations, and be extra careful with the {2N825 and the TLCU541! The circuit requires a separate power supply that provides regulated output voltages of 5 V (+ terminal) and 15 V (r+ terminal). The current requirement forthe 5-V supply is only 50 mA or so, while that of the 15 V supply is determined mainly by the collector current of the TUT. In most cases, 200 mA will be adequate Voltage regulators such as the 7805 and the 7815 are fine for these applications, but do not forget the usual decoupling capacitors to prevent noise and oscilla- tion, The completed! PCB is fitted in a suit- able ABS enclosure, the size of which de- pends on whether the power supply is internal oF external In any case, do not use mains adapters to power the circuit, since these do not in general provide the required output voltage stability. ELEKTOR ELECTRONICS MAY 1990 The curve tracer is connected to the computer by a standard printer cable. ‘The curve tracer is simple to use: insert or connect the transistor under test (make sure you get the bce terminals right), apply power and run the control program by clicking twice on ‘npn.prg’ in the file menu. The program, after being loaded, will prompt you to enter the transistor typeand type any key to start plotting. Do not worry if nothing appears to happen at first, since the Iy = 0 WA curve is drawn first! Once the output characteristic ap- pears complete on the screen, the program halts and waits for a key to be pressed to take you back to the file menu. Hard copy may be obtained before exiting the pro- gram by disconnecting the tracer from the printer port, connecting the printer, switching it On line and pressing the aL TERNATE and HELP keys simultaneously. Finally, the circuit and the program are suitable for testing n-p-n transistors only. ‘The control program supplied on disk is suitable for monochrome Atari ST sys tems only. . Reference: 1 Transistor curve tracer, Elekfoy Electro- nies October 1988.INTRO-SCAN FOR CD PLAYERS J. Ruffell ‘Owners of inexpensive, yet reliable, CD players, such as the Philips CD371 or Aristona CD1372, are often frustrated by the limited programming facilities of these machines. This drawback becomes particularly annoying when a CD is to be copied on to a cassette. The program With most of the popular CD players, tis necessary, when 4 certain number of tracks of a CD are fo be copied in a particular onder on to a cassette, to place the track numbers into the memory. This in itself is not a big problem if you haye the CD box to hand. This is not always the case and can put an end to this irritation. the only solution is thew to listen to all the tracks and note down the relevant track numbers, This can, however, be a time: consuming business ‘The programming aid presented here makes it possible, with the aid of the ‘next” ey, to listen briefly to all the tracks described here fon the CD during which each track num. ber may be entered into the memory of the programming aid. After all tracks have been scanned, the content of the aid’s memory is transferred to. the internal memory of the CD player, Provision is made for the CD player to be started auto- oe Fig. 1. Cireuit diagram ofthe programming aid ELERTOR ELECTRONICS MAY 1990matically af taken pla ‘The board is smal) enough to be fitted inside the CD player: Connexions between itand the front panel are by a 10-way flat cable. Two key switches and an indicator LED must be addled to the player controls, Circuit description The circuit oF the Fig. tis based on IC>, a register with a capacity of lox bits that stores the Ps grammed track numbers When the unit is switched on, IC> is reset via C7, Ro and ICs, and the SR bistable formed by IC3< and ICxt Re, ICs and C21. The Q output (pine 8 oF ICA) of the bistable then disables the square-wave generator based oi [C2e Via Ruy, Cre and 12. The freq generator is determined by (Pir R7}-Cu When in this condition the ‘play’ key of the CD player ts pressed, the first ack of he CD is played. At the same counter ICéis increased by 1. Every time the ogramming aid—see eney of the the “next” key sed. briefly is pi counter position is increased by 1. Monos- table ICs key and tthe clock input of the counter, en sures that the pule train emanated by the next” key is converted into a singl The keys on most popular CD operate by tone decading, This means that continuous pulse trains exist at pins 2 and bannected hetween the “next 4 of KI. Pressing the ‘play’ or the “next” switch places the pulse train on pin 1 or pin 3. A pase train at pin 1 does not mat rer boca abling the counter, but that at pin 3 is.con verted into a single pulse by [Cu Each time the MEM switch, SI, is pressed, ICT necetves a clack pulse via TC3s position of IC is stored in the reg: ister. The relevant programmed. The LED will light when at least one number has been placed into the register. ‘Atter all track numbers (ap to 15) have been stored, the ‘stop’ key of the CD player muist be pressed. The display then shows the total number of tracks on the CD. The contents of IC the player's memory by simulation of the ‘manual programming of the player a pro- se the pulses are used only far en nck mimber is then are transferred, to cess that is started by pressing TRANSIER switch Sz. The bistable is then set, 50 that the counter receives a reset pulse via Ri- Aiter a brief delay caused by Ri-Cr, the square-wave generator is started. Each pulse emanating from the generator oper- ates the “next” switch; at the same time, the counter receives a clock pulse via ICs so that it remains synchronous with the track indication om the display The position af In is compared by IC with the first four-bit word stored in the register, Ifthe data are identical, the A=B. ‘itput af IC¥ emits a pulse that is passed to the programme key of the player via INTRO-SCAN FOR CD PLAYERS, 61] ICs and pins 5 and 6 af Ki. The relevant number is then stored in the play’ "The next stored number then appears at the output of the register, after which the generator sends as many clock pulses to the “next” switch and the counter as are necessary to make the counter position co. incide with that at outputs Qi-Qs of IC this manner, the circuit scans all the track numbers of the CD and emits a program pulse at the moment the relevant numb appears at the output of I which this happens depends on the setting ‘of Pl In principle, there are no limits 49 the speed so long as the copying takes place correctly Gates ICe1, ICs and ICxb-d ensure that the circuit is reset correctly when the high: test counter position is reached. Wire bridge A enables automatic start ing of the CD player when the contents of the register have been copied to the player When copying is complete, a pulse is sent to the ‘play’ switch, If this facility is not roquired, the bridge is simply omitted The speed at Construction The printed-cireuit board shown in Fig. 2 is not available reudy-made. It is, how ever, easy to make and once it has been populated, it is conveniently built into the CD player. The connexions from the board to the player's front panel controls, fare as shown in Fig. |. The supply fine g -ARTS LIST Resistors RI,A2,RiO.A1 = 100k R3=47k R4,F5,R6,AB = 1M. r= 8k6 Fg = 470k 2 = 3308 Pt = 50 preset Capaciors: 1m 242, 10 2,04,08,68,c8,09,012 = 1000 63=220p 7 = 4u7, 10 610-22), 10 1-474, 104 ‘Semiconductors DI =LED, red, 3 mm icy = 74HoTa06S loa. 7aHoTi4 (C3 = 74HCT0O ied =74HCTES ios 74HCT123 (05-4516 (NOT 74HOTES16)) 107 = 74HCT40105 Toa = 74HOTS2 Miscellaneous: Ki = 10 pole male header 1,82 = keyswitch, t make Fig. 2, Printed circuit board ofthe programming ai. ELEKTOR ELECTRONICS MAY 1990 sOcmot 10-way flat cableline, pins 9 and 10 of Ki, is connected to the relevant terminal on the CD board. The earth line, pins 7 and 8 of Ki, is com nected to the relevant terminal in the CD player, Pins 1 and 2,3and 4, and Sand 6 of Ki respectively, are connected to the terminals of the ‘play’ and store’ key switehes of the CD player eee respectively. All these connexions are made by a short length of 10-way flat cable. MEM switch St, TRANSHER switch 5: and the red LED should be housed on the player's front panel in a convenient position. Finally, verity that the aid works satisfactorily. One common fault is that the contents of the aid's memory are not transferred correctly to the player's memory. This is invariably caused by the square-wave generator ‘operating at too high a frequency. The remedy for this is setting Pi to a higher resistance value. . PROTECTED HIGH-SIDE DRIVER Designed for use as a general purpose, single-channel, high- side (sourcing) power driver. Sprague’s. UDN2901Z is a smart power IC that can fune- tionally replace p-n-p darling- ton power transistors jn many applications, ‘Over-curres been designed into the device and is actuated between 1.5 A and 24 A. It protects the de- | 4 vice from output short circuits with supply voltages up to 25. V. When the maximum protection has WPVM MYT LAO neant-pa mun linear voltage regulators or (with aan extemal ground clamp diode) as a relay/solenoid driver. Owing to the nature of the protective cir- cuitry, the device iy protected when ‘operating in either the linear condi- tion or a “saturated” mode (e.g. when driving relay/solenoid loads). ‘The device should NOT be used as a low-side driver (p-n-p emitter follower configuration), ‘The UDN2901Z is supplied in a Head JEDEC power-tab TO-220 plastic package for operation over a temperature range of -20 °C to driver output current is reached, the obt- put drive is reduced linearly. If the over- current condition continues, the thermal shutdown operates limiting the junction temperature. SOA protection (Wee 2 15 V) is provided by limiting peak cur- rent as a function of the voltage across the device. Though the device is p-n-p-like in its functional behaviour, it is actually. a composite p-n-p/n-p-n darlington with several notable differences, including in- creased current gain, reduced gain band- width, and increased input threshold voltage. The device will always draw some standby current owing to the eur rent requirement of the protection cir- cuitry, When the input is OFF. the pro: tection features are disabled, The UDN2901Z is intended for use as a high-side driver. Typical appl tions include use as a pass transistor in +85 °C. For automotive and industrial applications, the UDQ2901Z can be supplied for operation down to 10°C. Source Data sheet 29310,30 from Sprague, 115 Northeast Cutoff, Box 15036, Worces- ter, Mass. 01615-0036; (SO8) 853-5000 or Sprague Electric UK Ltd, Salbrook Road, Salfords, RHI SDZ; telephone (0293) 517878. ELECTRICAL CHARACTERISTICS at T, = +25°C, T,,, = +70°C, V, = HV. Ta Chart simbot_| Testondtions Ti — Tye — Wa | is Fantona Spay Range 7 pepe y Gaiscoreniin Beabiom Gage | iis | Eee ee or CotectorEmier Sauron tage | Vors =TRe=—ToRA =e as Lv 14K, > — toma eee oiecor Ena Stang Wage] Vans TALS onH ae Fives put Goren Teas | Wa= es 08 phe Standby Carer Tan Last CY TEVEEO. Se Waa asVe =O Sat ae aan Ge Te SOMA V TA a i= 500nA Vea ee ‘Current Limit ous ty SOMA, Veg = 14.V_ 55 20] -24] A i= Om Veto og} 0s [-oT A Sat Forward Core Tana Rato | fy [stave = 200 | 00 |] — Tera Sow t Sef e inal Tab Thana esaaree [Re a NOTE: Negative currentis defined as coming out of (sourcing) the speciied device pin. EKTOR ELECTRONICS MAY 1990LETTERS Levers of @ general nature, or ex- pressing an opinion, or concerning a matter of common interest in the field of electronics (in its widest sense), should be addressed to The Editor at our London offices. Their publication in Elekror Electronics is atthe discretion of the Editor. DANGEROUS “DESIGN IDEAS” Dear Sit—The proposal that “a thyristor. rated at not less than 25 A, is connected directly across the mains supply” (Febru- ‘ty 1990) is extremely unsound and any- ‘one experimenting with anything as nasty as the circuit in Fig. | invites damage and injury. What happens at switch-on, and switch-off? What about “spikes” on the ‘mains? Spurious triggering of the thyristor could cause entire positive cycles to be shorted out and there is no attempt to limit the current that might flow. Where are the fuses? ‘When you publish these flawed de- signs with descriptions of their operation, you inevitably make them credible ~ de- ‘spite the disclaimer. Your publication en- couriges experimentation and so you have aduly of care 19 exercise. Someone should vet these ideas for electrical safety before publication. Readers realize these circuits ‘may not work properly. Do they know that they may be lethal? The apparent disregard of danger spoils what is otherwise an imaginative idea although the zero-crossing region is, actually a tricky part of the cycle to re- cover data fron: zero-voltage switches on other equipment inject noise there. Good results can be obtained by superimposing frequency-modulated carriers on to the ‘mains via isolation transformer CM. Stake (Stake Consultants), Woking. The author comments: “Mr Stake is en tively right in commenting that the thyris- tor must nor be connected directly across the mains: 1 should have made elear in both rext and drawing that connection 10 the mains should be made at atl tn ‘an Isolaring transformer, anid that the cir- cuit should be protected by suitable fuses.” After looking into the design and arti- cle in detail, 1 would add thar it appears that both the author and aur consuliing staff have made the assumption that ‘it will be clear to the reader that ay isola ing transformer and suitable fuses should be used’. Fam in agreement with you that such assumptions should not be made TE} S232. OR V242 Dear Sir—1 have taken Etekror Blectron- es for some considerable time and am ‘most intrigued to find a magazine that is available throughout Europe using the North American term for the standard computer serial interface. The term is, of course, RS232 or RS232-C, whereas the European designation is V24. It is true that Recommended Standard 232 detines the pins for each function, the voltage levels at these pins and the type of connector, that is, 25-way D type, whereas the V24 specification only specifies the various in- terchange circuits’ functionality. However, the permitted voltage levels are contained in CCITT Recommendation V28. When it is considered that many home computers do not use a 25-way D type cconncetor for their serial communications port.it is clear that RS232-C is incorrect [would have thought that any publica- tion that sets out to educate its readers should at least attempt to use terminology that is current in its country of publication, B.G. Oldford, Co. Cork. You ave right, but unfortunately most per sonal computers and certainly all IBM PCS and compatibles, including those ‘manufactured in Europe, have RS232 01 RS252-C oullets, never V24 ones. If man- nfacturers were to sell PCS with a V24 tn terface, Lam sure very few customers would recognize this. So, sheer force of numbers and Americanttapanese influ ence, unforsunately, dictate the use of S232 [Ed] ELEKTOR ELECTRONICS MAY 1990