Australia Austria Belgium Denmark France Germany Netherlands New Zealand Norway Sweden Switzerland S 1.50' S. 40 F. 60 Kr. 10 F. 8 DM. 4.20 Of L. 3.50 $ 1 .50 Kr. 10 Kr.. 14 F r 4,40 recommended 45 January 1979 UK. 55 p. Can. $1.75 Computers and Chess AC millivoltmeter Universal digital meter How I beat the monster up-to-date electronics for la UK 4 — elektor January 1979 decoder elektor Volume 5 45 decoder Number 1 Elektor Publishers Ltd., Elektor House, 10 Longport, Canterbury CT1 1PE, Kent, U.K. Tel.: Canterbury (0227) 54430. Telex: 965504. Off ice hours: 8.30 - 12.46 and 13.30 - 16.45. Bank: 1. Midland Bank Ltd., Canterbury, A/C no. 11014587 Sorting code 40-16-1 1, Giro no. 3154524. Z U.S, A. only: Bank of America, e/o World Way Postal Center, P.D. Box 30689, Los Angeles, CA 90080, A/C no. 12350-04207. 3. Canada only: The Royal Bank of Canada, c/o Lockbox 1969, Postal Station A, Toronto, Ontario, M5W 1W9. A/C no. 160-269-7. Please make all cheques payable to Elektor Publishers Ltd. at the above address. Elektor is published monthly. Number 51/52 (July/ August! is a double issue. SUBSCRIPTIONS: Mrs. S. Barber Subscription 1979, January to December ind.: U.K. U.S*A./Can, other countries surface mail airmail surface mail airmail £8.50 $21.00 $31.00 £8.50 £14.00 Subscriptions normally run to December incl. Subscriptions from Frebuary issue: U.K. U.S. A. /Can. other countries surface mail airmail surface mail airmail £ 7,75 $ 19.00 $ 28.00 £ 7.75 £ 13,00 Back issues are available at original cover price. Change of address: Please allow at least six weeks for change of address. Include your old address, enclosing, if possible, an address label from a recent issue. ASSISTANT MANAGER: R.G. Knapp ADVERTISING MANAGER: N.M. Willis National advertising rates for the English-language edition of Elektor and international rates for advertising in the Dutch, French and German issues are available on request, EDITOR U.K, EDITORIAL STAFF W. van der Horst I. Meikleiohn J. Barend recti t G.H.K, Dam P. Holmes E. Krempelsauer G. Nachbar TECHNICAL EDITORIAL STAFF A. Nachtmann J, Oudelaar A.C. Pauptit K. S.M. Wal raven P. de Winter Technical telephone query service, Mondays only, 13,30 - 16,45, For written queries, letters should be addressed to dept. TO. Please enclose a stamped, addressed envelope or a self -addressed envelope plus an IRC, ART EDITOR: F. v. Rooij Letters should be addressed to the deportment concerned: TO “ Technical Queries ADV = Advertisements ED = Editorial (articles sub- ADM - Administration mitted for publications etc.) EPS - Elektor printed circuit SUB *= Subscriptions board service The circuits published are for domestic use only. The submission of designs or articles to Elektor implies permission to the publishers to alter and translate the text and design, and to use the contents »n other Elektor publications and activities. The publishers cannot guarantee to return any material submitted to them. All drawings, photographs, printed circuit boards and articles published in Elektor are copyright and may not be reproduced or imitated in whole or part without prior written permission of the publishers. Patent protection may exist in respect of circuits, devices, components etc. described in this magazine. The publishers do not accept responsibility for failing to identify such patent or other protection. Dutch edition: Elektutir B.V., Postbus 75, 6190 AB Back (L), the Netherlands. German edition: Elektor Verlag GmbH, 5133 Gangelt, W-Germany French edition: Elektor Sari, Le Doulieu, 59940 E&taires, France. Distribution in U.K.: Seymour Press Ltd., 334 Brixton Road, London SW9 7AG. Distribution in CANADA: Fordon and Gotch {Can.) Ltd., 55 York Street, Toronto, Ontario M5J 1S4- Copy right ©1978 Elektor publishers Ltd. — Canterbury. Printed in the UK. ■ IMTH Of TM £UDIt W CtfCuufHM What is a TUN? What is 10 n? What is the EPS service? What is the TQ service? What is a missing link? Semiconductor types Very often, a large number of equivalent semiconductors exist with different type numbers. For this reason, 'abbreviated' type numbers are used in Elektor wherever possible: •'741' stand for pA741 , LM741, MC641, MIC741, RM741 , SN72741 , etc, • 'TUP' or TUN' {Transistor, Universal, PNP or NPN respect- ively) stand f or any low f re- quency silicon transistor that meets the following specifi- L cations: UCEQ, max 20V 1C, max 100mA hfe, min 100 Ptot, max 100 mW fT, min 100 MHz Some TUN's are: BC107, BC108 and BC109 families; 2N3856A, 2 N 3859, 2 N 3860, 2N3904, 2N3947, 2N4124. Some 'TUP's are: BC177 and BC178 families; BC179 family with the possible exeption of BC1 59 and BC1 79; 2N2412, 2N3251 , 2N3906, 2N4126, 2N4291. • 'DUS' or 'DUG' v Diode Univer- sal, Silicon or Germanium respectively) stands for any diode that meets the following specifications: DUS DUG EJr, max 1 F , ma x Ifl, max Ptot, max C P, max 25V 100mA VA 250mW 5pF 20V 35mA 100 MA 250 mW 10pF Some 'DUS's are: BA127, BA21 7, BA218, BA221 , BA222, BA31 7, BA318, BAX13, BAY61 , 1N914, 1N4148. Some J DUG's are: OA85, QA91 , OA95, AA116, • 'BC107B', 'BC237B', '8C547B' all refer to the same 'family' of almost identical better-quality si I i co n tra ns i stors. I n genera I , any other member of the same family can be used instead. 8C107 <-8, -9) families: BC107 (-8,-91,80147 (-8, -9), BC207 {-8, -9), BC237 (-8, -9), BC31? (-8, -9), BC347 1 - 8 , -9), BC547 (-8,-9), BC171 (-2, -3), BC182 (-3, 4), BC382 {-3,4), BC437 (-B, -9), BC414 SCI 77 {-8, -9) families: BC177 (-8, -9), BC157 {-8,-9), BC204 (-5, -6), BC307 (-8, -9), BC320 M,-2), BC350 (4,-2), BC557 (-B, -9), BC251 (-2,-3), BC212 (-3, 4), BC512 (-3,4), BC261 (-2, -3), BC416. Resistor end capacitor values When giving component values, decimal points and large numbers of zeros are avoided wherever possible. The decimal point is usually replaced by one of the following abbreviations: P (pico-l = 10' 13 n (nano-) = 10" 9 M (micro) = 10“ fl m (milli-) = 10" 3 k (kilo-) = 10 s M (mega-) * to 6 G (giga-) = to 9 A few examples: Resistance value 2k7: 2700 O. Resistance value 470: 470 n. Capacitance value 4p7: 4.7 pF, or 0,000000 000004 7 F . . . Capacitance value 1 0n; this is the international way of writing 10.000 pF or .01 p F, since 1 n is 10' 9 farads or 1000 pF. Resistors are % Watt 5% carbon types, unless otherwise specified. The DC working voltage of capacitors (other than electro- lytics) is normally assumed to be at least 60 V. As a rule of thumb, a safe value is usually approxi- mately twice the DC supply voltage. Test voltages The DC test voltages shown are measured with a 20 kf2/V instru- ment, unless otherwise specified, U, not V The international letter symbol 'U 1 for voltage Is often used instead of the ambiguous 'V', 'V' is normally reserved for Volts'. For instance: U b = 10 V, not V b = 10 V. Mains voltages No mains (power line) voltages are listed in Elektor circuits. It is assumed that our readers know what voltage is standard in their part of the world I Readers in countries that use 60 Hz should note that Elektor circuits are designed for 50 H 7 operation. This will not normally be a problem; however, in cases where the mains frequency is used for synchronisation some modifi- cation may be required. Technical services to readers • EPS service. Many Elektor articles include a lay-out for a printed circuit board. Some — but not all — of these boards are avail- able ready-etched and predr M led. The 'EPS prim service list 1 in the current issue always gives a com- plete list of available boards. • Technical queries. Members of the technical staff are available to answer technical queries (relating to articles published in Elektor) by telephone on Mondays from 14.00 to 16,30, Letters with technical queries should be addressed to: Dept. TQ. Please enclose a stamped, self addressed envelope; readers outside U.K, please enclose an IRC instead of stamps, • Missing link. Any important modifications to, additions to, improvements on or corrections in Elektor circuits are generally listed under the heading 'Missing Link' at the earliest opportunity. contents elektor January 1979 — UK 5 mmmS-ii xl. ■ ■ : >V W *-i*. Mt ■" ■; , v m i « ■ i ■ 1 1 . ■ ' I i i i ii i W-.: u • . k»bh p! ; ! ' ' ' | 1: i " , : 4S ir.!Kf|r jjl tT iif fc - u i m- - ?V ... .... J*S! ; 1 The American Institute of High Fidelity (IHF) recently published a new standard, dealing with 'Methods of Measurement for Audio Amplifiers'. When it comes to buying audio equipment, it would be a great help if manufacturers started 'Measuring by the book*. p. 1-02 A reliable nicad charger should prove a useful item to owners of port- able radios, flash guns and the like: the prices of nicad cells have dropped to the point where it has become a viable pro- position to use them in almost any type of battery-powered equip- ment. p. 1-08 How wilt a chess-playing computer fare against a strong human player? Mr. Levy descri bes how the reigning world cham- pion computer lost . . . 'How I beat the monster' should prove of interest! A companion article takes a look at the back- ground of Computers and chess. p. 1-34 ... - Universal digital meter How J beat the monger Digital displays are replacing conventional pointer instruments in many applications. Rapidly falling prices now seem to be hastening the final demise of the pointer instrument: a digital panel meter is now actually cheaper than its analog counterpart. AC mHHvoitmetep 5KS | » K— 1 contents selektor 1-01 measuring by the book 1-02 IHF to neb urs t generator . * . ■ . * 1-06 For certain audio measurements, the new IHF standard specifies a particular toneburst signal. reliable nicad charger . 1-08 improved LED VU/PPM(j.m. Heuss) 1-12 In April 1977 {Elektor 24) a design was published for an audio output level meter. With the aid of an add-on circuit it is possible to improve the resolution of the meter at the lower end of its scale. oscillographics on board . . , . , 1-13 The oscillosgraphics circuit published in September 1973 has proved to be a highly popular design, and several readers have asked for a printed circuit board, AC millivoltmeter and signal squirt 1-16 The lowest AC range on most multimeters is usually several volts If.s.d.) and It presents a relatively low load impedance to the circuit under test. A "preamplifier for multimeters" can solve this problem. DC polarity protection 1-19 sixteen logic levels on a scope ir. Rastattsri . . . 1-20 When trouble shooting digital circuits it is often useful to be able to examine the logic level of a number of signals simul- taneously. class tells 1-22 This article takes a look at a recent commercial power ampli- fier design which combines the inherently low distortion of Class-A output stages with the high efficiency of a Class-B configuration. ejektor 1-25 Electronically variable resistance, FM-stereo generator 1*26 Using relatively straightforward means it is possible to con- struct a simple yet extremely useful FM stereo generator, which can be employed to check the operation of stereo decoders and FM receivers* passive oscilloscope probe 1-30 The 10:1 probe described here can be constructed from standard parts, applikator 1-32 Programmable sound generator, computers and chess . 1-34 How the monster thinks. chess challenger 10 plays like a human 1-39 how I beat the monster (David Levy) 1-40 universal digital meter 1-43 Digital replacement for pointer instruments. automatic emergency lighting unit 1-48 market UK-15 advertiser's index UK-26 advertisment ■ 1 ™* ■ Elekfor January 1979 - UK 9 i mr face PoWCT SAiPP'V Keyboard vco moduli ,VCF 19724*1) . ■ ■ ■ iA0 SR 19725 1' . i Dual VCA 19726 1) ,LFO 19^7 1) - ■■ • NO'W , COM 1977 M ■ ■ ■ • front paiiflKc.pl • 3oki, Keyboard"^ K A contacts . CpI *• *"' h k 7 t l thus i 1 " 1 '*’ OA/^C n? * £50. — * all necewerv VCO and V * £365 — ADS Hi.. and Fr ° n ' 09 F " ELECTRONICS LOTS OF COMPONENTS AMO KITS LJ IUI Hi t’lCM t ii>, 1 1 l 5M ■ f L | ! I'JLjr i .' i". I ! | M*1| 1 7* Ul.- L ’4 Hl -T l IlJUiSn j LLjJ I o i LLil I 1 L TLM J4J v y mjiHUli i 71 \ I JOT i (I;* i L! JU L LJ J-l b n>Jsut i, L*AI In b L HA . Ml: t t, L B I L SI bib I l.'llli i LN/UIH. L 1 fr*z]ii i 1 JC- I r 1 < H r - V j 1 TAJUttl Y rCA2&J i TAAj¥ 1 T [AaUHi b Ul!llJb i iJtlU! iytt Y lSIklAJil b LH jaHihJhh t LH Iblljl i L jiJj j l-ij&h '- [JHi! Mil b IMIlinm i >1 II ?K Til I b Ul .1 E f M Tl '.'JH L LMUHJji U»lHfTOoJj L TUI ill L iJTd.'N I i r I .' A l Iff b."..l. I I 1“ v‘l I'M.' I ! . A I • . A 1 !Jl.bJ'J t IJi R-f Y U' >■>& LK bV I b LH J 1>5- 1 LHI? j I yiiri LK Mil> LH brit |.H bH IJI !Mn I ¥ ! H ? IJI HI 1,1 H !J. i A.U lu ■'bib I r i.A ^ I' lh BALhijg TAA^U i h.iVilJ Ibt’j ^3 It K 'i ',h SAHYhU lKjI? 7^:h ■ JbJi ■: JU I AM, I I H ri.iW.2WL ^ »A®J)b I AMn -■ H-! IHAHllw «A f Oj iiATi ■*.■.■! l .A .'I. ¥ I lHtJ. ilA i'll f*klj it I uA/U'iJ-j I uJkf LU LtHi4i u ■ LI d : L uATLirtMl* uA ?2 Sd ■ L iiA. ,- J iTUfafa ua TC jl?» uAV I '* 1 IlJiKii oA.«-.,J|. | uA J >i J I '.Hife <4>TiH>n Jl b AA^IIU ■ Aa -'ii ^tii i i : .\a I'n '.ru I hA ■’ SL Maw i ! llAH I ‘I HAH; i l^ifl- l -J I i TAAdn - 1 TV r AAd-^ft r, Ayii.i i i-Wb} IAAs*9JD TDftMMH T DA I UOh TUA I SJOi TD4IQQB riiAiaia H-. I v! I TK «*ul N UP Hit'J J J 2 P Hi.: I J-JsF nuaii^p Hci Jyjf LHL 4 '. Hm q Lhm':Hd.i . LhL^'jHTcj .1Vl4ftrilr TC-aZLOJI; V-AJu .! H '.A t IJ i f ; LA TlJhl « LA Jufll. LAJUhA'T IA5UB L-L UAJUtfb CA^LIHME -l'AAO^uai,' LA Hlv.ii ‘•■Hi HI9-4L '-HiiH £a 1| 5ut CATil4i.lt la } | N-J | 1-' i Hprc Lh ]'¥l.i i LH >“f l.i'i 1 |J|W|. | Li 4 | M> m.w |S | W* P I 4 ** ■ ll 1 ^ w- -I* IQIJIU : h KLJ lu? ICLfe- i !>H SN-'^^S I .SNrwi. msi 'pi, • W-MlHi Hh‘.. I j U LH : .*h r J 7»KillLL' i'Aj.XtJL ?MajKL ’¥ lx Jl ,'hhlKl ’HMiiKL .'■.ftiiL t jbiDL tjou CAN gy PHOWE fHoiA «ortD/»y riu feiOAy ti oo en To i?.oo pki SAT 0£E>AM iO.^o Aki to l3.co PtA At H iLLiwqrofj *i) LETTEEfo; MllCe HUTCHINSOKlj 2 CVNM Co»D. GKimstopJ, JCIMEJS t-yfJrJ, HORfoLK ft 32.1 AD, CMCOUE And Po^Tj'LoKW* ONty to THE. WAM£ oF "DG 8o£f^ £L6CTizor4|Cg . AuLPeiCGS A 22 VAT-|MCLUTJ£d rod sop rfee Post amd package 0MGCS6AS emotes Pt£AS£. Ttt HoU. AN p BElou) HO HOUSE*CAUS PlfA^e -P U£CObKoM flwr Jjiso LiuMiVAbT J-COwSoW^ 7 " -touch oMM« 5 r* / 1-fOWK j - Slcyc^ _„= £ ZMO I epfiutflTOft “ ' Mi ELEkTOR klTS Y* Gigahertz counter CPL (9^87) £ 104.50 A. Time base control iS&B?-!! £ 48.00 S. Low fr&tqueotie mput amp <9807 -3) 6-65 C- Countpr and displav <9887-2! £ 85,90 D. High frequently input Bnnp (9887-41 . £ 15.30 Automatic mono/jtpreo switch (99231 £ 6.30 T V. scHjnrt modulator 19925! £ 6.30 Mini counter (99271 £ 27.70 Mini short wave receiver (9920) £ 9.0S Digital reverberaiion main board (9913-1) £ 67.25 Digs tale reverberation extension board <991 3 21 £ G9,&5 Percolator swuch (9902! ... £ 8-40 Colour modulator 1.9873) ... £ ^4.00 Moving cml preamp (9911! . ■ £ 19.50 El r k to rn ado (w i lh ou t hyatsi n ks ) (98741 £ 20.— Colour T.V. oamei board 19892) £ 24.45 Intra red light gate transmittyr <9862-0 £ 3.30 Infra red lig^H gate receiver (9862-2! ........ £ 9.85 (jRvelopmyn l timer 190401 . . £ 18.50 Elektor equaliser <9032! . ... £ 23.— UAA 1BQ LED meter (9817 1+21 . £ 12-10 Simply function generator (9453 S £ 27,75 Signal injector (97&S! £ 6. — Sensi tive I ightmyter (9060) . . £ 12.55 M’.W. reflex receiver <9000! . . £ 7.55 Healing controller (9877! ... £ 2B.50 Analogue iryq. meter <9889! . £ 10.15 Elek try t mike preamp 1 98SS! . £ 5.10 UHF T.V modulator <9864) £ 3.70 Magnetise* 19827) ...... £ 4,50 Senior for Rlertmmeter (9826 2) £ 5.BO ElektromRter <9826-1 ! £ 5.00 Video tuo feedback generator (9B25-1) £ 12.35 Video bio alpha amplifier (9825-2) £ 12.35 loni«r (9323) . £ 11,40 infra red transmitter (9822) . £ 23.05 Log darkroom timar (9797) . C 10.05 F.M, mains intercom 19359) . £ 37.00 VA Digit OVM (77109! ... £ 23.50 4 Watt car radio amp <77101 ) £ 5.25 T. V, game* with AT-3-E500 177084) . . . £ 13-75 Guitar pryflmp (77020) £ 6-05 Precision time base 19448( £ 15.BS Power supply for pnyr.. time base (9443 1> £ 5,60 Micieofwcessoie sc/mp. Kir * HAM I/O (904&-1! f 32.35 * SC/MP hoard <9846-2) ... £ 26.75 * CPU card <9851 T £ 90,50 * BUS board (98571 £ 3.— * Memory card 19863!* £ 57.50 * HFX I/O <9893!* . . £ 67.25 * 4-k HAM (90051* £ 122.05 * Powms r su ppl y 19906)* . . £ 23.05 * Cass, interface (9905!* , . , , £ 16,05 * 3 £l bo q programmed E PR O Ms * £ SB. 95 *Cp! system IrnnsiRtR of kits witti "] £ 349 — el ecT&ouic - MUSIC - i ? i 6 f*% ‘ . *W*P (tHlAi % 5 AO pflij MfN) r ■* £H€QuE £)> us> PoStaL nOKBSL OfjLH TO THefJRM£ Of D€ JwSf FL&OTfSo^lOS. fat- o^e. Prices [,/r-!Nci0D£Z> ! P.Db sop roe pgp. > de boer f / elehtromha/ Ktekne Berg 39-41 Eindhoven. Nederland, t«k 040-448229 Elektor January 1979 - UK 14 advertisment From Science of Cambridge: the new MK 14. Simplest, most advanced, most flexible microcomputer - in kit form. PROM -51 2 bytes RAM -256 bytes. Extra RAM (optional) 4.43 MHz crystal 5 V regulator Power rails and input/output edge connector RAM I/O device (optional) CPU 8-digit, 7-segment LED display PROM-' 512 bytes RAM -256 bytes Extra RAM (optional) MK 14 including optional RAM I/O and Extra RAM. Display and keyboard interface circuitry. Edge connector for _ external keyboard with up to 32 keys The MK 14 is a complete microcomputer with a keyboard, a display, 8 x 512-byte pre- programmed PROMs, and a 256-byte RAM programmable through the keyboard. As such the MK 14 can handle dozens of user-written programs through the hexadecimal keyboard. . Yet in kit form, the MK 14 costs only £39,95 (+£3.20 VAT, and p&p). More memory— and peripherals! Optional extras include: 1. Extra RAM -2 56 bytes. 2. 16-tine RAM I/O device (allowed for on the PCS) giving further 128 bytes of RAM. 3. Low-cost cassette interface module - which means you can use ordinary' tape cassettes/ recorder for storage of data and programs. 4. Revised monitor, to get the most from the cassette interface module. It consists of 2 replacement PROMs, pre-programmed with sub -routines for the interface, offset calculations and single step, and single- operation data entry. 5. PROM programmer and blank PROMs to set up your own pre-programmed dedicated applications. All are available now to owners ofMK14. A valuable tool— and a training aid As a computer, it handles operations of all types - from complex games to d igital alarm clock functioning, from basic maths to a pulse d elay chain. Programs are in the Manual, together with instructions for creating your own genuinely valuable programs. And, of cou rse, its a superb education and training aid- providing an ideal introduction to computer technology. SPECIFICATIONS •Hexadecimal keyboard • 8-digit, 7-segment LED display • 8 x5l2 PROM, c on tain ing monitor program and interface instructions •256 bytes of RAM • 4 MHz crystal • 5 V regulator® Single 8 V power supply® Space available for extra 256-byte RAM and 16 port I/O® Edge connector access to all data lines and I/O ports Free Manual Every MK14 kit includes a Manual which deals with procedures from soldering techniques to interfacing with complex external equipment. It includes 20 sample programs including math routines (square root, etc), digital alarm clock, single-step, music box, mastermind and moon landing games, self-replication, general purpose sequencing, etc. Designed lor fast, easy assembly The MK 14 can be assembled by anyone w ith a fine-tip soldering iron and a few hours' spare time, using the illustrated step-by-step instructions provided. How to get your MK 14 Getting yourMK 14 kit U easy, Just fill in the coupon below;, and post it to us today, with a cheque orPO mad e payable to Science of Cambridge And, of cou rse, it comes to you with a comprehensive guarantee. If for any reason, you're not completely satisfied with yourMK 14, return it to us within 14 days for a full cash refund. Science of Cambridge Ltd, 6 Kings Parade, Cambridge, Lambs., CR2 1SN. Telephone: Cambridge (6223) 31 1488 n Tor Science of Cambridge Ltd, 6 Kings Parade, Cambridge, Cambs., CB2 1SN. Please send me the following, plus details of other peripherals: □ MK 14 Standard Microcomputer Kit U £ 43.55 (inc 40p p&p.) 0 Extra RAM mS 50 Dm 6 SOOrna 79031 2 resistance. If desired, of course, all four switches contained in the IC may be connected in parallel. When the switches are closed, the sinewave attenuation is determined by the ratio of R14 to R19 (for AC, the lower end of R19 is effectively grounded!). When the switches open, however, P2 and R18 are connected in series with R14, To calibrate the unit, PI should first be adjusted until the beginning and end of each burst coincide with the zero- crossings of the sinewave, as described in the original article. P2 is then set so that the level difference between 'burst on’ and 'burst off’ corresponds to 20 dB (x 10). H Lit: Tone-burst generator, Elektor December 1978, p. 12-10. measuring by the book elektor January 1079 — 1-07 IHF Standard therefore specifies the A - Weigh teds igna l- to -n o ise ra tio ( S /N) . This is the ratio of the output reference level to the A -weighted output noise level, in dB. The measurement is per- formed on all inputs, with the gain 4 3 3 On 3 3 On 04 soon hpl |“p| Y o 101 o T T TpEh f S 47 n rlh-O : 0 ♦ A-weighting filter ■ ^ 7903G 4 control set according to standard test conditions (e.g. 0 dB overall gain for a line input). The input under test must be terminated with a specified im- pedance: a Ik resistor for line inputs: a 100 £2 resistor for moving-coil inputs; the network shown in figure 5 for dynamic disc inputs. No nonsense: no short-circuited inputs. The terminations specified are reasonable approximations of the type of input load that will normally be found in practice, and so the measurement results are likely to be more realistic than in the past, A great idea. The standard does include a warning: care should be taken that the termin- ation networks do not act as pickup links for electrostatic or electromagnetic fields. Not to worry: any unwanted pickup of this type would produce poorer figures — and no manufacturer will make that mistake. A similar measurement can be performed using the CCIR weighting network. In this case, the result is listed as the l CCIR/ARM signal-to-noise ratio'. Last but no least, the Transient overload recovery time of an amplifier refers to the time required for an amplifier to recover from a 10 dB overload of 20 ms duration, occurring at a repetition rate of once every 0.5 seconds. The tone-burst signal shown in figure 1 is applied to the relevant input. The gain control is set so that a reference-level input will produce an output level 10 dB below the continuous average powder rating (or voltage output rating, as the case may be). The output signal is monitored on an oscilloscope; in par- ticular, that portion of the cycle that occurs immediately following the return of the input signal to reference level. The number of sinewaves that are still visibly distorted is determined; this figure, expressed in milliseconds, is the recovery' time. In conclusion It will probably take a while for the new Standard to penetrate into advertising copy. Since it originates in the USA, it is to be expected that American (and Japanese) manufacturers will be the first to use it. What about Europe? Embarrassed silence. The Common Market and European Unity notwithstanding, Decibels made easy The decibel (dB) is the tenth part of a ‘Bel', but for some reason decibels are the only 'unit 1 ever used. Who ever heard of cent ib els or millibels? The dB is used to specify voltage, current and power ratios (to name a few) — on the lines of: how much larger is this voltage with respect to that one. The number of dBs is defined as 10 or 20 times the logar- ithm of the ratio in question; TO limes' for power ratios and £ 20 times' for voltage or current ratios. Why? In particular, why use logar- ithms, and why use a different multi- plication factor for power ratios? To start with the first point: the ratios that we are interested in can be very large, easily going up to 100,000:1. Furthermore, in most cases where dBs are used, ‘significant 1 changes in level are those where a signal is multiplied by a certain factor. For instance, the perceived increase in a power level from, say, 10 W to 20 W (x 2) is the same as the perceived increase from 100 W to 200 W! In both cases the power is doubled, and that is what counts. Expressed in dB, both ratios are equivalent to a 3 dB increase. Use of the dB also solves the first point: power ratios from 100,000:1 to 1:100,000 correspond to a dB scale from +50 dB to -50 dB (for power ratios). In spite of this ‘scale compression’ a (significant) ratio of 2:1 is still clearly expressed as +3 dB. The second point, the difference in multiplication ratios, is easy to explain. Power (in Watts) corresponds to voltage squared divided by resist- ance, Therefore, if a voltage increases from, say, IV to 3 V (x 3), the corresponding power will increase from, say, 1 W to 9 W (x 9). However, the effect is the same: these are just two different ways of expressing the same level change. When using dBs it is effects that interest us, and so it is preferrable to use the same dB-value to express both ratios; in other words, if the power ratio is given in European manufacturers and standards institutes still haven’t succeeded in coming up with a similar up-dated- standard. They’re working on it, give them their due, but there's a distinct lag compared to the situation in the States. Come on boys, hurry it up! DIN 45.500 is dead. It's high time it was buried. N Lit: The official IHF Standard document IHF- A - 2 02 can be obtained from : The Institute of High Fidelity, Inc. 4S9 Fifth avenue. New York N. Y. 10017 , USA . Price : $ 7.50. dB, it is useful to give the voltage ratio squared . Since dBs are logar- ithmic, this is equivalent to multi- plying the logarithm of the ratio by 2: if ‘power-dBs’ are 10~x the log of the ratio, voltage (and current, etc.) dBs must be 20 x the log of the ratio. So much for theory. In practice, dBs now lead a life of their own. Although it is quite possible to calcu- late logarithms of ratios, there is usually no need. Bearing in mind that adding dBs is equivalent to multi- plying the ratio, only a few dB values need to be memorised in order to ‘calculate’ virtually any ratio with a sufficient degree of accuracy. Here we go: number power voltage ratio of dBs ratio ratio 0 1:1 1:1 +3 2:1 \/2:l (=sl.4: +6 4:1 2:1 + 10 10:1 \/T0: 1 («s 3: +20 100:1 10:1 To give a few examples: A voltage ratio is specified as 32 dB. 32 = 20 + 6 + 6, so the ratio is 10x2x2 = 40. Voltage ratio 34 dB: 34 = 20 + 20 - 6, so the ratio is 10 x 10 x Vi = 50. Note that the minus sign simply implies that the ratio works ‘in the opposite direction': +6 dB = 2:1 (= x2), so -6 dB = 1:2 (= Voltage ratio 33 dB, Now what? Using a ‘rule of thumb’: 1 dB is approximately equal to 10%' for volt- ages (20% for power ratios), so 33 = 32 + 1 (or 34 - 1) and the ratio is 40 + 10%** 44 (or 45). The output power of an amplifier is 60 W. How many dBW? In other words, what is the power ratio in dB between 60 W and reference level (1 W)? One approach: 60 W is ‘just over’ 50 W; 50 = 1 00+ 2, so 50 W is 20-3= 17 dBW. 60 W must be a bit more ... How much more? 1 dB^ 20% - perfect! So 60 W must be 18 dBW r A spot-on accurate calcu- lation would give the following result: 17.78 1 5 1250 ... T8’ is near enough. 1-08 — elektor January 1879 reliable mead charger reliable nicad charger Nickel-Cadmium accumulators, or nicad cells for short, are becoming ever more popular. As prices drop and the number of available types increases, it has become a viable proposition to use them in almost any type of battery-powered equipment. For this reason it is reasonable to assume that many owners of portable radios, cassette decks, electronic flash guns, pocket calculators and the like — not to mention remote-control enthusiasts - are on the look-out for a really good nicad charger. It shouldn't be expensive (otherwise it's still cheaper to use conventional batteries), but it must automatically provide the correct charging current and time: on the one hand, overcharging must be impossible; on the other hand, the cells must be fully charged. Anyone who has had previous experi- ence with charging nicad cells will probably have discovered what features are desirable for a nicad charger. For that matter, even without the benefit of past experience - good or bad — the basic requirements are fairly obvious, A good charger should be reliable, and under no circumstances should it damage the nicad cells. Regrettably, not all commercially available chargers fulfil these requirements. The charger described in this article was designed to meet the following specifi- cations; • it should be suitable for practically all commercially available types of nicad cells; • the charging current should be held constant, at 1/10 of the capacity of the accumulator in Ah; however, it should also be possible to select a higher current for so-called sintered cells, since these may be charged at 1/3 of their capacity; • a timer should be incorporated in the charger, to ensure the correct charg- ing time ; • to preclude the possibility of damag- ing the nicad cells, they should first be discharged to a well-defined level before starting the charging cycle. In this way, the danger of drastically overcharging near-full batteries is virtually eliminated; • preferably, the changeover from discharge cycle to charging cycle should be carried out automatically; • after completing the charging cycle, it should be permissible to leave the cells connected to the charger {for months, even). Furthermore, under these conditions they should be trickle-charged to keep them fully- charged at all times. Phis list of requirements was presented to one of our designers, with the request to come up with a cheap and reliable circuit, suitable for home construction — an important point, if it is to be published in a magazine! — that would do the job properly. After the usual head-scratching, breadboarding (more of a bird's nest, actually), testing and evaluating, a circuit evolved. The underlying principles are best clarified with the aid of a block diagram. Block diagram Figure 1, the functional block diagram of the nicad charger, illustrates the basic principles of the final design* Compared with most conventional chargers, this block diagram may seem frighteningly complicated. In practice, it is not nearly as bad as it seems, since several of the 'blocks' actually consist of quite simple units. For instance, the block marked Trickl e-charge* merely represents one resistor. To start at the beginning; When the start pushbutton is operated, a flip-flop (FFl) initiates the discharge cycle for the nicad cell(s). Ibis is indicated by a red LED. When the voltage across the cell (or cells) drops below a preset level, a comparator (IC1) resets the flip-flop FFl. As a result, the discharge cycle is terminated and a second flip-flop (FF2) is triggered. The charge cycle is now automatically initiated; a green LED lights to reassure any passing nicad owners . , . When FF2 changes state, a timer is triggered* This unit is included to fulfil one of the main requirements stated earlier: the charging cycle must be terminated automatically after a certain time has elapsed. At the end of the preselected time, FF2 is reset by the timer. The main charger (Tl) is cut off, but the trickle-charger remains operative, maintaining the cells in the highly desirable fully-charged state. The charging current can be adapted easily to suit most of the commonly available types of nicad accumulators. The ‘heaviest 1 cells that can be charged are 1,2 Ah types. However, when de- signing a circuit and, more particularly, when specifying component values, some basic assumptions must be made. In this case, the component values specified are valid for charging the most commonly available 0.5 Ah types of nicad cell. These cells are charged at 1/10 of their Ah rating - 50 in A — for the specified period of 1 4 hours. When charging sintered- cells - as stated earlier, these cells can withstand charging at up to 1/3 of their Ah rating — the charging current can be 'upped' to 150 mA; at the same time, the charging time is re- duced to Vh hours. Wide-awake readers may have noticed that both these reliable mead charger elektor January 1979 — 1-09 Figure 1. Block diagram of the automatic charger for mead accumulators. The cal! {or cells) is first discharged to a preset level, after which the fixed-duration charging cycle is initiated. Figure 2. The complete circuit. Fairly stan- dard components are used throughout. FF1 r FF2 = 102 = 4013 M2,N4_ N6 = IC4 = 4093 N1,N3 = JC5 = 4023 79024 2 1-10 — elektor jamiary 1979 reliable nicad charger periods exceed the nominal charge of the cells; however, they may rest as- sured: the total charge is well within the manufacturer’s tolerance. Charging with constant current has a significant advantage in practice: it makes no difference whether a single mead cell or a series-connection of up to 6 cells arc charged at one time. Complete circuit The block diagram shown in figure 1 is derived from the complete circuit given in figure 2, The only additional com- ponents in the final circuit are the main supply (mains transformer, bridge recti' fier and capacitor C 5 ) and two switches, S3 and S4, which offer the possibility of 'manual override': they can be used to initiate and terminate the charging cycle. Since the basic principles have been explained above, the discussion of the complete circuit can be relatively brief. S2 is the 'start' button. Operating this pushbutton sets FF1 the Q-output of this flip-flop goes ‘high’, T3 and T2 are turned on, discharging the nicad cell(s) via a L fat' resistor, R7. Simultaneously, D2 lights. After a certain time, the ceil(s) will be discharged to the point where the voltage across it {or them) drops below the level preset by means of PI. The latter voltage and the voltage across the cell(s) are applied to the two inputs of a comparator, IC1. Normally speaking, the voltage across a ‘fully’ discharged nicad cell is taken to be approximately 1 V. For the discharge cycle to be terminated at the correct point, the voltage preset by PI should be set at the number of series^connected nicad cells times 1 volt. Assuming that PI has been set correctly, the output of the comparator will switch from 'high’ to ‘low’ when the cell(s) are fully discharged. Via Nl, flip-flop FF1 is reset; its Q output goes 'low’, turning off T3 and T2 and ter- minating the discharge cycle. Simul- taneously, a differentiating network (C2/R12) resets FF2 via N2. The Q output of this flip-flop goes high, turning on T4; as a result, the current source (Tl) comes into play. The nicad cells are charged, and D1 lights. Note that a green LED must be used for D1 : this diode is not only used as indicator, it also provides the reference voltage for the current source. The voltage drop across green LEDs is higher than that for red LEDs (2.4 V as opposed to 1,6 V) and the specified values for R1 and R2 are only accurate for the higher voltage. When FF2 changes state, initiating the charging cycle, its Q output goes 4 low\ This enables the timer. The timer circuit is the essence of simplicity: it consists of a clock generator - N5 , N6, an inverter incorporated in IC3 and a few passive components - and a frequency divider (1C3). The frequency of the clock generator can be adjusted, by means of P 2 , until the correct liming intervals are obtained. Only a few components in the circuit remain to be discussed. R13 andC3 are included to reset the two flip-flops Parts list Resistors: R1 = 33 n R2 = 10 n R3 = 2k2 R4 h R9,R12 - 10 k R5 “Ik R6 = 1 20 n R7 = 10 m/ 5 watt R8 = 390 a R1 Q r R 1 1 ,R 1 3 - 22 k R14 = 10 M R1 5 = 3M9 Capacitors: Cl = 10^/16 V C2 = 1 r>5 C3 = 4/u7/l6 V C4 * 560 n C5 = 1000 m/16 V Semiconductors: Tl = 8D 140/BD 136 T2 = BD 139/BD 135 T3 h T 4 = BC 547 D1 = LED green D2 = LED red SCI - 741 IC2 = CD 401 3 IC3 = CD 4060 1C4 - CD 4093 ICS = CD 4023 Sundries: Pi = 1 k preset F2 - 1 M preset SI = double-pole, double throw S2 h S3 f S 4 ~ single-pole pushbutton B = bridge rectifier B40C800 F = 1 00 mA fuse Tr = 9 V/25G mA mains trans- former reliable nicad charger elektor January 1979 — 1-1 1 rl I '' IF JE S2 79024 - 4 Figure 3, Printed circuit board and component layout for the nicad charger (EPS 79024). The only component mounted 'off-board' i$ the mains transformer. Figure 4. This wiring diagram clearly illustrates the connections from the p.c. board to the various switches, the transformer and the nicad cetl(s). when power is initially applied — the circuit simply works, without first having to fiddle with all sorts of reset buttons. Resistor R3, tucked away in the top right-hand comer of the circuit, is the ‘trickle-charger’ : even after XI has cut off it continues to supply a small charging current into the nicad cells, in order to keep them ‘topped up\ Finally, the switches. The charging current is selected by means of Sla; with the values given for Rl and R2, the current is 50 mA in position 1 of this switch and 150 mA in position 2. To avoid mishaps, a second pole of the same switch selects the corresponding charging time: the two positions of Sib correspond to 14 hours and 3 Vi hours, as mentioned earlier. Normal operation is initiated by operating the start button, $2; as explained, this actually initiates the discharge cycle. If one is in a hurry, operating S3 initiates the charge cycle without first discharging the cells. At all times, both the charge and discharge cycles can be stopped by operating S4. Construction and operation The design for a printed circuit board and the corresponding component lay- out is shown in figure 4. The connec- tions to the external components — mains transformer, switches and nicad cell(s) — are given in figure 5. Basic construction is fairly straightfor- ward, therefore; initial calibration and normal operation is hardly more com- plicated, There are only a few points to watch: • As stated, from 1 to 6 cells (connec- ted in series) can be charged at a time, provided PI is set correctly: 1 volt per cell. Note however, that D2 will not light during the initial discharge cycle if only one cell is connected. Furthermore, if more than one cell is to be charged they should all initially be discharged to approximately the same degree; if they have ail been used in the same item of equipment, this will normally be the case. In case of doubt, it is advisable to first dis- charge each cell (or set of cells from the same unit) individually until the green LED just lights. • If other charging currents are re- quired, the values of Rl and/or R2 must be modified accordingly. The charging current (in amps) is equal to 1.6 V (the voltage drop across D1 minus the voltage drop across the base-emitter junction of T 1 ) divided by the value of Rl or R2. If the unit is to be used for rapid charging of 1.2 Ah nicad cells (charging current approximately 360 mA), T1 should be provided with a cooling fin, • If desired, accurate calibration of the timer circuit can be carried out by means of P2, However, there is no real harm in merely setting it in the mid-position. , . Perfectionists may consider this rather less than satis- factory, whereas they may also be rather reluctant to sit out the com- plete timing cycle of 3Vi or even 14 hours. No problem: there is yet another alternative. Monitor the Q4 output of IC3 (pin 7) with a multi- meter and operate the start button (no nicad cells connected). If this out- put swings positive after 45 . . . 50 seconds, P2 is set correctly. * One final - important - point: since this circuit first discharges and then charges the cell(s), no protection diodes could be incorporated at the output. Care should therefore be taken never to connect the cell(s) the wrong way round; furthermore, if they are left connected to the charger after the charging cycle has been completed, the charger must remain switched on. Otherwise the cell(s) would be Xrickle-d/^charged 1 through R3, R5 and PI 3 H 1-12 — elektor januarv 1979 improved LED VU/PPM improved LED VU/PPM In April 1977 (Elektor 24) a design was published for an audio output level meter. The meter, which incorporated a LED 'thermometer-scale' display, could be modified to give either a 'VU' or 'PPM' type of response. With the aid of the following add-on circuit, it is possible to improve the resolution of the meter at the lower end of its scale, allowing much more accurate measurement of signal levels during quiet passages of music and speech. J.IVJ. Heuss The original meter employed a column of 20 LEDs to display the output level of each channel. As far as the upper half of the display was concerned, the differ- ence in signal level between two success- ive LEDs was 1 dB; the lower half of the display, however, was scaled in steps of 5 dB, In conjunction with the relatively long discharge time of storage capacitor C4 (see figure 2 of the original article), this meant that the display was unable to register small and rapid variations in low level input signals. This is illustrated in figure la of this article, w T here curve a represents the rectified audio input signal and curve b is the voltage on C4, i.e, the voltage which is actually displayed on the LEDs, However, if one arranges that, as soon as the voltage on C4 drops below a preset reference voltage U re f the capacitor’s decay time constant is reduced, the volt- age across this capacitor will track rapid variations in the rectified input voltage much more closely. This process is I illustrated in figure lb, and in fact represents the basic function of the circuit described here. The additional components required to improve the meter’s response are shown within the dotted lines in the circuit diagram of figure 2, This diagram is for one channel only and should of course be duplicated for stereo applications. Comparator Kll (KIT), the potential divider resistors and LED are as shown in figure 5 of the original article. The actual operation of the circuit is straightforward. When the voltage on C4 (which is applied to the non-inverting input of Kll) falls below U re f , LED D3 1 is extinguished and transistor T2 is turned off. T3 is then turned hard on and R83 appears in parallel with C4. This reduces the discharge time constant of C4, with the result that, at low input levels, the voltage across this capacitor follows fluctuations in the rectified audio signal much more accurately. If the "peak memory’ switch SI is in- cluded, it should be fitted with a second pole, Sib (for stereo applications two extra poles will obviously be required). It is worth pointing out that the fre- quency response and the "ballistics’ (attack-decay times) of the meter can be improved by replacing the 741 op-amps with more modern and faster devices. For example, if less than unity gain is required, an LF356 can be used for IC1, If a gain greater than 5 is desired, then an LF357 is also suitable (see original article regarding the choice of values for R2 and R3). As far as IC2, IC3 and IC4 are concerned, an LF356 will prove eminently suitable. If an LF356 is also used for IC5, then the offset adjustment can be dropped (i.e. potentiometer P2 omitted). oscillographics on board elektor january 1979 — 1-13 oscillographics on board The Oscillographics circuit published in September 1978 has proved to be a highly popular design, and several readers have asked for a printed circuit board. For the benefit of those who have not seen the original article, the basic principles of the circuit are repeated here in brief. Although an oscilloscope is an ex- tremely useful instrument, it spends most of its life ‘displaying’ a horizontal line or even a blank screen. The Oscillo- graphics generator, shown in figure 1 , can remedy this: it produces a multi* tude of fascinating and attractive geometrical patterns on the screen of the scope. The patterns are actually so-called ‘Lissajous’ figures: two res- onant circuits (IC2/IC3 and IC5/IC6) are triggered at regular intervals by a multivibrator (IC1), producing two damped sinusoidal outputs. These are fed to the X- and Y -inputs of the scope, producing an intriguing display. Both the frequency and the decay rate of each sinusoidal output are indepen- dently variable (by means of P 1 /P3 and P2/P4, respectively), so that a virtually infinite number of different patterns can be obtained. An 'intensity' output is provided {Z or Z, depending on the type of scope), which can be used to blank the spot on the screen during triggering of the resonant filters. It is also possible to modulate either (or both) oscillator signalfs) by an external signal applied to the M x and My inputs, so that the patterns are continuously changing. Various types of (low fre- quency) modulation signal can be used: square wave, triangle, ramp, etc., with varying amplitude and frequency. The only constraint upon the modulation signals is that they should not contain a DC component (in other words, they should be AC coupled), since otherwise there is the possibility that part of the pattern will be off the screen. The maximum amplitude of the modulation signal is 15 Vpp. If desired, the values of R13 and R2Q (and/or R14 and R21) can be altered if the effect of the modulation signal is more or less notice- able than was intended. Power supply A suitable supply circuit for the Oscillo- graphics generator is given in figure 2. The positive supply Tail is stabilised by an IC regulator (IC9). The negative rail is referenced to the positive rail by means of an opamp (IC 10) and a transistor (T2). This is an interesting little circuit, which can prove useful in many other applications: the voltage on the negative rail is maintained at such a level that the voltage at the R24/R25 junction is 0 V, Since R24 is equal to R25, the negative output voltage must be equal and opposite to the positive output voltage. In other words, if the positive voltage is varied, the negative voltage will vary in step — providing a variable symmetrical output voltage. However, in this application a fixed ± 5 V supply is required. The printed circuit board Although it is the raison d'etre for this article, the p.c, board (figure 3) requires little comment. It accommodates the Oscillographics generator and the power supply, with the exception of the mains transformer. _ Both Z and Z modulation outputs are provided. These are, of course, only useful if a modulation input is provided on the oscilloscope. Depending on the type of scope, one or other of these outputs will give the desired result. If the picture is not completely flicker- free (again, this depends on the charac- teristics of the scope), then the value of Cl can be reduced. 1-14 — elektor January 1979 oscillograph ics on hoard Figure 1. Circuit diagram of the Gscillo* graphics generator. Figure 2. A symmetrical stabilised power supply is also incorporated on the p.c. board. Figure 3. The printed circuit board for the Oscillograph ics generator (EPS 9970). 4016 4066 © ILL 3 - % Pn r r^l - ! t1 T _L iHAHAHd- G oscillographics on board elektor January 1979 1-15 Parts list Resisto rs: Capacitors: Sem iconductors; R1,R9 r R1G,Rl6,R17,R24 r Cl = 47 n T1 = BC 1 07, BC 647 or equ. R25 - 10 k C2 . . . C© = 10 n T2 = BC 1 77,BC557 or equ. R2,R3„R8,R1 1 r R 1 2 P R1 4 r Rt 5, C6 r C7 = 220 m/16 V IC1 . , . IC7,)C1 0 = 741 R 1 8,R1 9,R21 - 100 k C8,C9 = IO/i/IO V 108 - 4016,4066 R4 = 22 k IC9 = 78L05A(C)Z R5 r R23 - 4k7 01. . . . 04 = 1N4001 R6,R22 - 1 k Sundries: R7 = 2k2 Trl =2x6 V/100 mA mains R1 3/R20 - 220 k transformer PI . 10 k lin. Si = double-pole mains switch FI = 100 mA fuse 1-16 — elektof January 1979 AC millivolt mater and signal squirt AC millivoltmeter and signal squirt It is often useful to be able to measure low-level audio signals. However, the lowest AC range on most multimeters is usually several volts (f.s.d.) and — to make matters worse — it presents a relatively low load impedance to the circuit under test. A 'preamplifier for multimeters' can solve this problem. Since a suitable preamp should have a high input impedance and a sufficiently large bandwidth, it seems logical to use FET-input opamps, Furthermore, since a single 1C contains four of these opamps it is a relatively simple matter to include a 'signal squirt' on the same board. Figure 1. Clock diagram of the millivoltmeter (upper section) and of the signal squirt (lower section). Figure 2, The average value of a full-wave rectified sine wave is higher than that of a h< If -wave rectified signal. Furthermore, for the full- wave rectified sine wave , the RMS value is 1*1 1 x the average value. Figure 3. Complete circuit diagram. As in the block diagram, the upper section is the millivoltmeter and the lower section is the signal squirt. Figure 4, For calibration purposes, the R1/C1 junction must be offset by 45 mV (DC). This can be achieved by temporarily adding Ra and Rb, as shown. The circuit described here is quite useful in its own right. The output is not only suitable for driving a multimeter — a normal panel meter can be used instead. It is also possible to use it in conjunc- tion with the ‘Universal Digital Meter’ described elsewhere in this issue. Fur- thermore, the current consumption is so low that a 9 V battery supply can be used, thereby retaining the flexibility and portability of the multimeter. When measuring very small AC voltages, it is important to ensure that the measuring instrument does not present an excessive load to the circuit being tested. This requirement can be fulfilled quite easily by using FET-input opamps, A further requirement is that the frequency of the signal being measured has no effect on the measured value. This is only possible, of course, over a limited bandwidth; therefore, this re- quirement can best be stated in two parts: the frequency response should be ‘flat’ within the specified bandwidth, and the bandwidth should be as large as possible. Based on these primary re- quirements, and bearing various second- ary requirements in mind (price, re- liability, availability of components), it was decided to use the T exas Instruments IC type TL084 - a quad FET-input opamp. The upper section of figure 1 is the block diagram of the AC millivoltmeter. An input capacitor blocks any un- wanted DC voltages, after which the remaining AC signal can be amplified to a reasonable level. So far so good, but amplifying the signal is not enough. A normal panel meter, or a multimeter switched to its most sensitive DC voltage or current range, tends to display the average value of the applied voltage or current. For a symmetrical AC signal, the average value is 0 V. . . In order to obtain a display, some kind of rectification is required. Even after amplification, the level obtained in this circuit is not so high that a simple diode AC millivoltmtiter and signal squirt elektor January 1979 — 1-17 man. SO mV I I □ A | I I T ri - [>CHl iy H3 4V7 400mWi Al . . A4=HC1 =TLG&4 Rt5_ a*2 + I I 0 _ | — | 50yA ♦ -0“ ' 2k BC547B 7903 S 3 as in series with the meter will suffice - the forward voltage drop across the diode would swamp the signal. Since there are plenty of op amps available in the same IC, the solution is to incorpor- ate the diode(s) and the meter in the feedback loop around an opamp — the forward voltage drop across the diode is then compensated for automatically. If a job's worth doing, it's worth doing well: instead of using half-wave rectifi- cation, as in most multimeters, full-wave rectification is used here. One advantage is apparent from figure 2. As stated earlier, the meter will measure the average value of an applied voltage. The average value for an AC voltage is zero; for a half-wave rectified voltage it is positive, so it can be measured; however, for a full-wave rectified AC voltage it is larger still — giving a higher reading on the meter. Extra ‘gain' for free! While we're on the subject of AC voltages, a further point is worth mentioning. Although the meter displays the average value, it is more common to specify the RMS {or 'effective') value of an AC signal. This is done to keep electronics simple. . . For DC, power (P) in Watts is equal to the voltage, squared, divided by resistance. For the same formula to be valid for AC, the RMS value of the AC voltage must be used. The abbreviation 'RMS' stands for ‘Root Mean Square', which is exactly what it is: the root out of the sum of the average values (mean) of the squares of the momentary voltages. This may sound quite complicated, but for the moment the only important thing to know is that the RMS value of a sinewave is equal to 0.707 x the peak value. The meter will display the average value, and for a sinewave this is 0.636 x the peak value. The ratio between RMS and average values is therefore 1,11, and when calibrating the meter the scale will have to be offset by this factor. This is no problem, as will be seen. So far, only two opampshave been used; however, the IC contains four. It seems a good idea to use the remaining two for a simple signal squirt - another item of test equipment that can prove quite useful to the home constructor. The lower section of the block diagram (figure 1) refers to this 'signal squirt'. It is little more than a conventional ‘Wien bridge’ oscillator. The Wien bridge proper is shown as a selective filter in the feedback loop around one opamp. Provided the total loop gain is greater than unity (no problem with opamps), the circuit will oscillate. The sinusoidal output is amplified by the last remain- ing opamp, to provide a 2 Vpp sinewave at the output. In order to obtain a ‘clean' sinewave, the loop gain around the first opamp should be almost exactly unity. To avoid critical cali- bration procedures, this adjustment is performed automatically: the output signal is rectified and fed back to a suitable control point. 1-18 — etektor january 1979 AC milhvoltmeter and signal squirt s Parts list: Resistors: R 1 6 = 82 k Semiconductors: R1 = 1 M R18 - 15 k 01 . . . D6 = DUS R2 - 6k8 R19 - 470 n 07 = 4V7/400 mW zener diode R3 = 68 k PI — Ik T1 ,T2 = BC 107B, BC547Bor R4 = 150 k P2ab - 47 k (stereo) equ. R5 = 150 k IC1 - A1 . . . A4= TL 084 R6.R7 - 47 k Capacitors: R8,R10,R1 1 = 100 n CUC6 = 1 jut R9,R17 = 1 k C2 = 100 ju/10 V R1 2 ~ 4k7 C3,C4 = 82 n Sundries: R13 = 100 k C6 = 100 W4 V 50 juA . . , 1 mA moving coil R14 = 100 k C7 = 560 n instrument or R15.= 8k2 C8 = 22 ju/16 V multimetar AC mi llivolt meter circuit The complete circuit is shown in fig- ure 3; the upper half is the AC milii- voltmeter. A reference voltage is derived from the 9 V battery by means of R8, D7, R9 and C 2!, This voltage is applied, via R1 and R2, to both inputs of the first opamp, A). R1 determines the input impedance (approx, 1 M). The gain of A1 is determined by the ratio of R3 to R2 — or, to be more precise, by the ratio (R3 + R2); R2 in this ease, a gain of x 1 1 is obtained. The output from A1 is fed to A 2 via a preset potentiometer (P 1 ); the latter can be used for full-scale calibration. The n on -in verting input of A2 is connected to the reference voltage across C 2 and a further (small) offset compensation is introduced via R6, R7 and P3. Four diodes, D3...D6, provide full- wave rectification. This part of the circuit operates as follows. Under quiescent conditions (i.e. without any signal applied to the input) the output of A1 will be equal to the reference Figure 5. Printed circuit board and component layout for the complete unit (EPS 79035). voltage. Via PI this voltage appears at the inverting input of A2; since the voltage at the non-inverting input is also equal to the reference voltage, the output of A 2 will be at the same level - any offset can be compensated by adjusting P3, When an AC signal is applied, the output of A1 will start to swing alternately positive and negative with respect to the reference voltage. When this voltage swings positive, the output of A 2 will swing negative - drawing current through D3, the meter, D6 and PI. Since the non-inverting input of A 2 remains at the reference voltage level, this op amp will attempt to maintain the same voltage at its in- verting input; in other words, the voltage drop across PI must be equal to the shift at the output of A1 caused by the AC signal. The current through PI (and the meter!) must therefore be proportional to the AC voltage — in spite of the diodes! When the AC signal at the output of A 1 swings negative, the same result is obtained — the only difference being that current now flows from the output of A2 instead of into it. In both cases, however, the current flows through the meter in the same direction - the diodes see to that. The final result is that an AC voltage applied to the input of the circuit produces an exactly proportional full-wave rectified current through the meter. The 'signal squirt' The basic principles of the sinewave oscillator have already been explained. A Wien bridge, consisting of RIO, Rll, P2, C3 and C4, is used as a highly selective filter in the feedback loop around A3. The resonance frequency can be set by means of P2. AC millivolt meter and signal squirt DC polarity protection elektor January 1 979 — 1-19 DC polarity protection The output from A3 is amplified by means of A4 (approximately xll). Diodes D1 and D2 rectify the output signal , charging 02; as the voltage across this capacitor rises, T2 is gradually turned on, this, in turn, causes T1 to be turned off; effectively, its 'resistance 7 increases. Since the ‘resistance 7 of T1 (in conjunction with R12) determines the gain of A3, this causes the gain to be reduced. Effectively, then, if the output voltage tends to rise above a certain level the gain is reduced; if the output tries to drop below that level the gam will be increased. There is only one option left for the circuit: the output level must remain constant. And it does. The output frequency can be adjusted from 500 Hz to 25 kHz (by means of P2). The output levei is approximately 2 Vpp, and protection against inadver- tent short circuits is provided by the simplest means possible: a resistor, R19. Calibration procedure As far as the signal squirt is concerned, calibration couldn’t be simpler: there isn’t any. The millivoltmeter is only slightly more complicated. First, the input to the circuit is shorted and P3 is adjusted until the meter reads exactly 0 V, Now, a DC calibration voltage of 45 mV is applied to the Rl/Cl junction. This voltage can be derived from the refer- ence voltage by temporarily adding resistors Ra and Rb as shown in figure 4 + Full scale deflection of the meter can now be adjusted by means of PL Since a DC reference of 45 mV is used, the full-scale deflection on AC will corre- spond to 50 mV — remember that fac- tor 1.11 between average and RMS values! Final notes A suitable printed circuit board and component layout are shown in figure 5 . The circuit can be used to drive any DC meter with a sensitivity from 50 juA f.s.d. to 1 mA f.s.d. — in other words, most commonly available panel meters as well as most multimeters. The maxi- mum input voltage is specified as 50 mV, although in practice it will normally handle voltages of up to 100 mV without difficulty. Higher voltage ranges can, of course, be included by adding suitable voltage divider circuits at the input. On the other hand, a multimeter offers the possibility of obtaining a more sensitive instrument: if the circuit is calibrated according to the above procedure on, say, the 500 /iA range, then a 1 00 fiA range will correspond to 10 mV f.s.d. The frequency response of the meter is ‘3 dB down 7 at 1 Hz and 125 kHz - in other words, the meter reads 30% low at those frequencies. Of greater interest for practical use, perhaps, is the fact that the reading (for a sinewave input!) is within 5% from 3 Hz to 40 kHz. N Electronic equipment which is fed from an external DC voltage can easily be damaged if the terminals of the supply are inadvertently transposed. In circuits which have only a small current consumption this danger can be averted by connecting a diode in series with the supply line. The diode will then only conduct if the supply voltage is of the correct polarity. If the diode is replaced by a bridge rectifier, then it no longer matters which way round the terminals are connected. However, particularly in circuits with larger current consumptions, this approach is somewhat unsatisfactory, since it leads to noticeable power losses. A more elegant solution, which results in no voltage loss and virtually no power loss, and hence is suitable for circuits carrying relatively large currents, is shown in the accompanying diagram. The component values were chosen for a DC supply of 1 2 V, The circuit should be mounted inside the equipment it is meant to protect and the external supply voltage connec- ted to terminals 1 and 2, Assuming the polarity of the supply is correct, once the on/off switch, SI, is closed, the relay, Re, will pull in, causing two things to happen. The normally closed contact, rel, will open, reducing the relay cur- rent through Rl. Since the drop-out current is less than the pull-in current, assuming Rl is the correct value, relay Re will remain energised. This little trick reduces the dissipation in the pro- tection circuit. Secondly, the normally open contact, re 2, will close, thereby applying power to the rest of the equipment. However, if the terminals of the supply are transposed, diode D 1 will be reverse- biased, preventing the relay from being pulled in. Diode D2 suppresses any inductive voltages produced when the relay coil is de-energised + If there is a fuse in the supply line of the equipment, then it is recommended that tills be inserted between the supply and the protection circuit, so that it will blow should a fault occur in the latter. The current consumption of the protec- tion circuit is so small compared with that of the equipment it guards that there is no need to alter the rating of the fuse. The values of the components in the cir- cuit can of course be modified to suit other supply voltages. One should bear in mind that the pull-in voltage of the relay, Re, should be the same as the supply voltage. The value of R 1 will depend to a certain extent on the type of relay used, and is best determined experimentally. N 1-20 — elektor January 1979 sixteen logic levels an a scope sixteen logic levels on a scope R, Rastetter When troubleshooting digital circuits it is often useful to be able to examine the logic level of a number of signals simultaneously. For example, one might wish to look at the logic state of all the pins of a dual-in- line 1C. To this end there are a number of commercially available test clips, which, via rows of LEDs, indicate which pins are at logic '1 If one has access to an oscillo- scope, a similar result can be obtained with the aid of the circuit described here. Figure 1. The logic states at the pins of the iC under test are displayed on the scope as shown here. Logic '0's are represented by a 'blip' appearing where the filled-in circles are drawn, and logic '1's by a blip where the dotted circles are. This pattern corresponds to the DlL-configuration of the IC pins. Figure 2. Block diagram of the Dl L-indicator. The circuit can be connected to the X- and Y- inputs of a conventional scope. Figure 3. Complete circuit diagram. The circuit is constructed using TTL ICs, although in principle it is also possible to employ CMOS. The DIL-indicator is used in conjunc- tion with a 16-pin test clip, which is attached to the IC being examined. From the sixteen input signals provided by the test clip the circuit generates two new signals, namely the X- and Y- input signals for the scope. Figure 1 illustrates how the logic levels are displayed on the scope screen: i.e, in the same pattern as the pins of the IC. The "high 1 logic levels are displayed slightly higher on the screen than the low 1 logic states, both being represented by a white blip 1 . The block diagram of the DIL-indicator circuit is shown in figure 2. With the aid of the two d at a-se lectors the sixteen logic signals on the test clip are scanned one at a time. Open inputs, for instance when testing ! 4-pin ICs, are represented as being at logic ‘1\ Both data selectors are clocked by a four-bit binary coun- ter, which in turn is controlled by a separate dock generator. The counter also clocks a digital-analogue converter, the output of which provides a staircase waveform with 8 'steps 1 . The staircase voltage forms the X-input of the scope and determines the horizontal position of each spot on the screen. The vertical position of the spots are controlled by the least significant bit of the counter and by the logic state of the signal selected by the data-selector, i.e. the Y-inpul of the scope is obtained by summing these last two signals. The complete circuit diagram of the DIL-indicator is given in figure 3, 1’he circuit is constructed using TTL ICs and is intended for use with this logic family. The two data- selectors of figure 2 are formed by 7415 Ts, whilst a 7493 is used for the four-bit counter. The square wave generator which provides the clock pulse for the counter is built up from two Schmitt triggers. The frequency of the clock oscillator is in the region of 70 kHz, By depressing switch SI this frequency can be lowered by around 3 kHz. This facility is needed lest the clock frequency of the circuit under test happens to coincide with that of the indicator circuit, with the result that a varying voltage may well appear constant. Operating the switch allows one to test for such an eventuality. The summing circuit of the block diagram consists of nothing more than three resistors (R7, R8, R9) f whilst the digital-analogue converter is scarcely more complicated; it consists of three NAND gates (N1 .... N3) connected as inverters and resistors R1 . . . . R6. The identical circuit can also be built using CMOS ICs, in the event that it is to be used to test CMOS circuits opera- ting off a voltage supply other than the 5 V used here. Although it is possible to do so in principle, it is not recommended that the indicator be built with the CMOS ICs and then used to test TTL circuits. (Problems may occur with correct timing and triggering). The 5 V supply (stabilised) should be capable of providing at least 125 mA. sixteen logic levels on a scope elektor January 1979 — 1-2t DOOOOQOO m "* OODDOOOO ■N £ data selector ^ data selector JTJ1_ \o rrn wtn/iH- 0 > counter O/A * ■ T T I 1 o tRna/nH $ 79003 3 16 15 14 13 12 11 1 C 9 DDQ0D0D0 S? testclip £ 3 E £2 DOOaDDOO B 5V © 16 12 13 14 15 □ 7 VCC □ 6 A Uo B D4 >ca 74151 M 02 C D1 Strobe 00 GND 1 U 11 10 15 14 13 12 9 10 11 Si sC TiLI m ... N4 - 7400 * IC4 N5 ... N6 = 7413 = IC5 R 9 c H 6 ^ 12 ca| W1Q ^ mm — [330 n I — J I— 1 f > _9j B 11 A ■ In A in B ' IC3 7493 *£G ON D Rq Ro 10 5 V O 70002— J H© class tells It would be nice to design a power amplifier which combined the inherently low distortion of Class-A output stages with the high efficiency of a Class-B configuration. The following article takes a look at a recent commercial design which appears to offer the best of these two worlds. Figure 1, Block diagram of a balanced push- pull output stage. The size of the current I (with respect to i) determines whether it is a Class-A or a Class-B type. Figure 2. By employing a combination of Class-A and Class-B output stages it is possible to obtain both the low dissipation of the latter and the excellent signal-handling characteristics How distortion) of the former. The Japanese company Technics has recently introduced a new stereo power amplifier, the SE-A1, which has an output of 350 Watts per channel. This in itself is nothing special, however the fact is that, despite its high output power, the SE-A1 does not employ a Class-B output stage. Although the price of the amplifier is such (around four thousand dollars) that it is beyond the means of most hi-fi enthusiasts, the operating principle of this new hybrid will doubtless be of interest to many readers. Class- A v. Class-B The debate on the relative merits and demerits of Class-A and Class-B power amplifiers has been waged for some considerable time now, and the argu- ments for and against are well-known: the primary advantage of Class-A amplifiers is low distortion, whilst their major drawbacks are price (more pounds per Watt) and their inefficiency. Class-B amps on the other hand are much more efficient, hence capable of providing greater output powers, but have higher distortion and are often claimed to sound worse than their Class-A rivals. The reasons for these differences in performance can be explained quite simply. In Class-B amplifiers the output transistor is biased to the cut-off point, so that it is conducting only over one half of the signal waveform (in practice the transistor is actually biased just above the cut-off point with the result that it conducts for slightly more than half a cycle). This means that in the absence of an input signal the quiescent current consumption, and hence power dissipation, is theoretically zero. Since only half of the signal waveform is passed by the output transistor, Class-B output stages use two transistors ar- ranged in a push-pull configuration. Each transistor conducts for half a period, so that current flows in suc- cessive half-cycles. This results in a considerable improvement in the ampli- fier’s efficiency. Unfortunately, ho w r ever, since the tran- sistors are biased to the cut-off point, it means that they cannot conduct fully until the base-emitter junction of the transistor is forward-biased by at least 600 mV. The transfer characteristic of the transistor is therefore non-linear near the cut-off point, causing what is known as "crossover distortion' during the transitional periods when one transistor is being switched off and the other is being switched on. Furthermore, since this distortion is relatively con- stant for any input signal level, the relative distortion of the output stage deteriorates at lower signal levels. In a Class-A output stage, on the other hand, the transistors are biased to conduct over the entire cycle of the input waveform. The output stage is thus considerably more linear than is the case with Class-B configurations, thereby reducing the amount of nega- tive feedback required to keep distor- tion to acceptable levels. Unfortunately, the other side of the coin is a consider- able reduction in the efficiency of the output stage due to the fact that a large amount of power is dissipated even under quiescent conditions. With the aid of judiciously applied negative feedback, the distortion levels of Class-B amplifiers have been reduced to levels which, to all intents and purposes, have no audible effect. However, if it were possible to virtually eliminate the inherent problems of Class-A output stages (inefficiency, low output powers), then such a modified Class-A system would represent an extremely attractive proposition. Class differences Before examining the modified Class-A design, it is worth briefly considering the conventional configuration of push- pull output stages. The triangle in the block diagram of figure 1 represents the driver stage, whilst the complementary output transistors are depicted as the two blocks marked P and N , in the absence of an input signal, a quiescent current I flows from the positive rail ■f U via P and N to the negative rail - U. When an input signal is applied, however, the collector current of one of the transistors will rise by a value i, whilst the collector current of the other transistor will fall by a corresponding class tells elektor January 1 979 — 1-23 1 u amount. The difference between the two collector currents flows through the load. Thus the greater i is, the greater the voltage developed across the load. The size of the quiescent current, I, determines which class the amplifier belongs to. In a class-A output stage 1 is sufficiently large that both output transistors are conducting regardless of the value of i. in a Class-B amplifier, however, the quiescent current is so small that, when a signal is applied to the bases of the output transistors, thereby driving them in antiphase, one of the transistors will soon be turned off, leaving the other to feed current into the load. A compromise between the above two types of output stage is the Class-A B amplifier in which the quiescent current is chosen such that, below a certain output power the output stage func- tions as a Class-A amplifier, (i.e. both output transistors are conducting), whilst above that point it operates as a Class-B amplifier (i.e. the output transis- tors conduct in turn). Class A + B To the above three types of amplifier a fourth can now be added: the Class A+ B, of which the Technics SE-A1 is an example. The block diagram of this new amplifier is shown in figure 2 . As can be seen, two amplifiers per channel are used: a Class-A amplifier and a Class-B amplifier, each with its own driver stage. The two output stages are fed from symmetrical supplies: The Class-B stage is connected to + and ~U 2 , whilst the Class-A stage is fed by the floating supply ± Ui t The junction of the Uj supply lines is connected to the output of the Class-B amplifier. The quiescent current of the Class-A output stage is I . If we now arrange that both amplifiers have the same gain (by choosing suitable values for Ri . , t R4), then for equal input voltages they must obviously have the same output voltage. Since the output of the Class-B stage feeds into the junction of the U ] supply rails, the supply voltages for the Class-A stage will follow variations in the output voltage. The result is that, regardless of the input drive voltage, the voltage developed across the two output transistors of the Class-A stage will always be (almost) the same (Uj). This being the case, Uj can be as small as is desired, in fact it need only be large enough to ensure that the output stage is still functioning satisfac- torily. This low value of Uj is the secret of the circuit. It means that the Ciass-A output stage consumes very little power (since the latter is the product of total supply voltage and quiescent current). The Class-A output stage itself delivers very little power into the load, since the AC voltage across each half of the output stage is almost zero. Under the influence of the drive voltage, the current through the Class-A output transistors may vary as much as it likes class tells 1-24 — elektor january 1979 (and does in fact do so), but the stage itself cannot be used to deliver output power into the loudspeaker. Any power developed is dissipated as heat; however thanks to the low supply voltage, this is limited to safe levels. Thus the task of actually supplying the Watts is left to the Class-B stage, which as we have seen, is inherently more efficient than Class-A output stages. As is apparent from figure 3 (figure 3a shows the case for a positive-going signal, figure 3b for a negative-going signal), the output current of the Class-B stage is fed to the load via the supply lines ±Uj and the Class-A output transistors. The current I is always sufficiently large to ensure that, regardless of the size of the AC signal current from the Class-B stage, the Class-A output transistors are never cut-off or saturated. Since the Classe-B amplifer has virtually no quiescent current, it is inevitably subject to crossover distortion. Does this then mean that the Class A + B amp must also be afflicted with this problem? Fortunately the answer is no. Although, in the absence of quiescent current, the Class-B output does indeed produce crossover distortion, the latter never actually has any effect upon the final output. Quite apart from the fact that the local negative feedback (round R3/R4 in figure 2) reduces the distor- tion to very low levels anyway, the sole effect of the Class-B induced crossover distortion is to cause a small difference in the output voltages of the two (Class-A and Class-B) stages - i.e, a small AC voltage is produced across the Class-A stage. The current source/drain characteristic of the Class-A output stage is easily good enough to ensure that this has a negligible effect upon the output signal. The result of employing both a Class-A and Class-B output stage in the above- described fashion is to obtain the high efficiency (disregarding the relatively small constant dissipation in the Class-A stage) and output power of the latter with the low distortion and excellent lin- earity of the former. The accompanying table lists and compares the power handling characteristics of Class-A, Class-B and Class A + B amplifiers for (an admittedly unrealistic) output power of 350 W into 4 £2, The distinctive performance of the A + B amp is expressed not so much in the figures for efficiency, but rather for maximum dissipation, which of course also has a decisive influence on the size of the amplifier and weight of the heal sink. In conclusion it can be said that the Class-A + B principle represents an interesting approach and one which need not always cost four thousand dollars to implement! H Figure 3. Current flow in the output stages of figure two for a positive and negative -going drive signal {figures 3a and 3b, respectively l 7901© 3 a 3b Table Power handling characteristics of Class-A, Class-B and Class A + B amplifiers. Maximum output power 350 W into load resistance of 4 H, A 8 A + B supply voltage Uj 53 V — 5 V supply voltage Uj — 53 V 53 V quiescent current A 6.6 A — 6,6 A quiescent current B — 0 A 0 A quiescent conditions power taken from supply 700 W 0 w 66 W dissipation 700 W 0 W 66 W output power 0 w 0 w 0 w efficiency 0 % — - 0 % output level at which power taken from supply 700 W* 284 W +66 W 35Q w maximum +6GW 2 Q 8 yy dissipation dissi pation 700 W* 142 W occurs + ow 14 ? w output power 0 w* 142 W efficiency 0%* 50 % 31 % max. output level power taken from supply 700 W 446 W +66W Rnw dissi pation 350 W 96 W +66W 162 W output power 350 W 350 W + o w 350 W efficiency 50 % 78% 68 % * Maximum dissipation occurs under quiescent conditions - ejektor etektor January 1979 — 1-25 LfU 'AliP electronically variable resistance For control of signal levels, particularly in tow-frequency circuits, some kind of elec- tronically variable resistance is often required. In the past, LDRs have been used in conjunction with LEDs or filament lamps; FETs or even bipolar transistors have been tried with varying success; one might even consider using a thermistor with some form of heating element. What these and similar approaches lack in sophistication they make up for in associated problems: distortion, noise, non-linearity, etc. Not to mention the difficulties involved in ensuring reasonable tracking between several units. However, the alternative proposed here seems quite promising. an invitation to invest igate, improve on and implement imperfect but interesting ideas Figure 1 shows a resistor, R1 ; a second resistor, R2, can be connected in parallel by closing a switch, S. If S re- mains open, the total resistance R between the two ends of the circuit is equal to Rl (see figure 2b); if S is kept closed, the total resistance is equal to RI//R2 (figure 2b). Not exactly world- shattering news, so far. However. If S is replaced by an elec- tronic switch; if this switch is operated at a high switching rate (well over the highest frequency the circuit is expected to handle); if the duty-cycle of the switch -control signal is proportional to some external control signal — then the complete circuit will operate as an electronically variable resistor! Why? Let us examine figure 3 , It is assumed that the control signal for the electronic switch is a squarewave with duty 50 dB (1 kHz!; 40 dB 110 kHz) stereo FM: >40 dB 11 kHz);30dB (10 kHz) encoder max. input voltage: 100 mV RMS frequency range: 30 Hz - 1 5 kHz distortion: <0.1% (1kHz) o$c, frequency: SB ... 108 MHz (variable) trans- output power: —6 , . . 0 dB miner max, frequency deviation: approx, 1 MHz harmonic suppression: > 50 dB for all harmonics elektor January 1979 - 1 27 FM -stereo generator Design As one might expect, there are several possible ways of generating an MPX signal such as that shown in figure 1. The L + R signal and pre-emphasis are simple enough to realise, however the modulated L — R signal presents slightly more of a problem. In addition, there is the separate question of suppressing the 38kIIz subcarrier and all signal com- ponents with a frequency greater than 53 kHz. In many stereo encoders (including that published in May *77) the L — R signal is obtained simply by inverting the R signal and then summing it with the L signal, before modulating the result onto the 38 kHz subcarrier. The circuit Figure 1. Frequency spectrum of a multiplex encoded stereo signal. Figure 2. Block stereo generator. diagram of the complete described here, however, operates on a different principle, which is both simpler and ensures better suppression of the subcarrier. The block diagram of the cir- cuit is shown in figure 2, The left and right channel audio signals are first pre-emphasised before being fed to achopper circuit which is driven by a 3 8 kHz signal. The latter is derived by dividing by two the output of a 76 kHz oscillator. The L and R signals are in fact sampled by opposite half cycles of the 38 kHz clock signal. Summing the L and R samples produces an MPX sig- nal which does not have to be separated from the 38 kHz subcarrier, since the latter is not in fact present - all that remains are the sidebands. The 19 kHz Amplitude 1-28 — elektor January 1979 FM -stereo generator 12 V (?) C14 47y MOD ♦see text 1 .“°f 1* — n 470 p L r P4- b IC1 - ESI ... ES4 = AOm D1„D2 ■ 1N4148 72,74 = 50516 * 71 ,73„75 = BO 549C or equ TG r T7 = BF 494orequ. 9959 3 pilot tone, which is similarly derived by frequency division (and lowpass fil- tering) from the 38 kHz subcarrier, is introduced via the summing amplifier. Before the MPX signal is fed to the modulator circuit, it is necessary to first feed it through a lowpass filter to eliminate all signal components above 53 kHz, The FM generator is simply an oscillator whose output frequency can be varied between 88 and 108 MHz, The MPX stereo signal is frequency modu- lated onto the oscillator to produce the FM stereo test signal which can then be fed via coaxial cable (or twin feeder) to the receiver under test. If only the stereo decoder in the receiver is to be checked, then the modulator and pre- ceding low pass filter are not required and the MPX signal can be fed direct to the decoder. Circuit As was apparent from the block dia- gram, the practical realisation of the cir- cuit is a fairly simple affair, since many of the functional blocks are contained in a single IC. For example, the four electronic switches used in the chopper circuit are present in one 4066, whilst the 76 kHz oscillator and the divide-by- two stages for the 38 kHz and 19 kHz signals are available in the shape of the well-known stereo decoder IC, MC 131 OP. The circuit diagram of the stereo multi- plex encoder is shown in figure 3, whilst figure 4 shows the circuit of the FM oscillator. The various sections of the block diagram can be easily identified in the circuit of figure 3, The pre-emphasis for the audio signals is provided by T1/T2 for the left channel and T3/T4 for the right channel. The feedback net- works R5/C4 (L channel) and R10/C6 (R channel) ensure that the gain of these amplifier stages increases above approx. 3 kHz, whilst below this fre- quency the gain is roughly unity. The maximum gain is determined by the ratio of R5 to R6 and RIO to R 1 1 ; C3/R5 and C7/R1 0 limit the bandwidth of the audio signal to approx, 30 kHz by rolling off the gain above this figure, IC2, the MC 1 3 1 OP, contains the 76 kHz oscillator and the dividers for the 38 and 19 kHz signals. The 38 kHz signal, which is available at pins 4 and 5 (the Q and Q outputs) of the IC, control the electronic switches ESI ... ES4, which sample the L and R audio signals. Since four of these switches arc con- tained in a single 4066, two switches per function are employed to improve switching efficiency: when ESI is closed, thereby letting the left channel signal through, ES3 is also closed, so that the right channel signal is simultane- ously shorted to earth; similarly, when ES4 closes, passing the R signal, ES2 also closes grounding the L signal. The sampled L and R signals are fed via R20 and R21 to the summing amplifier formed by T6 and T7. After extensive lowpass filtering (R31, L3, L4, L5, Cl 7 , . . C20, R34), the 19 kHz pilot tone at pin 1 0 of IC2 is also fed to the base of T6. The MPX signal can be taken (via Cl 2 and P2) direct from the output of the summing stage, however if it is to be modulated onto an FM car- rier, it must first be filtered to remove all signal components with a frequency greater than 53 kHz, This filter, which is formed by LI, L2, C13 and Cl 5, is absolutely necessary, since FM receivers will tend not to like the unfiltered sig- nal, and may react in rather an alarming way. The only section of the circuit which remains to he discussed is T5. Its func- tion is to forestall any problems which might arise between the MC 131 OP and the 4066 t Since the output level of the MC 131 OP is not fixed (it can be c.g, 4 or 6 V), some sort of buffer stage should be necessary. However, this would de- stroy the symmetry of the circuit, and for this reason T5 was included to pro- vide the supply voltage for the 4066, using the output voltage of the MC 1 3 1 OP as a reference. FM -stereo generator elektor january 1979 — 1-29 &3 L7 : 6 turns 1 mm dia. enamel ted copper wire on Amidon ring core T 50-12 L8 : 4 turns. 0.2 mm enamelled copper wire on ferrite bead L9 : 3 turns 1 mm enamelled copper wire wound on 6 mm dia. former; no core LID: 4 turns 0.2 mm enamelled copper wire on ferrite bead, tapped at 2nd turn FM test generator Figure 4 shows the circuit diagram of the oscillator which, when frequency modulated by the MPX signal, will pro- vide an FM stereo-encoded signal which can be fed direct to the antenna input of the receiver to be tested. Despite the fact that it is constructed using only one dual-gate MOSFET, even under varying load conditions the oscil- lator -is extremely stable, and exhibits very little temperature drift. The output filter (C28, C29, C30, L9) ensures good suppression of harmonics and also has the advantage of requiring no adjust- ment. The ferrite bead transformer at the output (L 1 0) allows connection to the receiver to be made either via coaxial cable (50 or 75 £2) or twin feeder. A ring core is used for the oscillator coil (L7), which means a high Q, and also has the advantage of reducing the like- lihood of either producing, or picking up r.f. radiation. The oscillator fre- quency can be tuned between approx, 88 and 108 MHz by means of trimmer C24. Construction and alignment Most of the components should prove to be fairly readily obtainable. Problems may be encountered, however, with the Darlington transistors, T2 and T4. if that is the case, then they can be re- placed by conventional BC 559Cs, pro- viding that R4 and R9 are reduced to 22 k. The chokes, LI . . . L6 should also be widely available, whilst as can be gathered from the details given in fig- ure 4, winding L7 . . . L10 should be a straightforward affair. The Amidon ring core can be obtained from T.M.P. Elec- tronic Supplies, Leeswood, Mold,Clwyd. The circuit as a whole is not especially critical, and assuming one takes the usual care, no difficulties should be encountered during its construction. Wiring should of course be kept as short as possible, particularly around ESI ... ES4, the input of the summing amplifier (T6/T7), and the FM oscil- lator. Thanks to the low current consumption of the circuit (approx. 40 mA), the power supply can be kept both simple and compact. The most obvious solution is to use one of the many common regu- lator ICs, as does the circuit shown in figure 5. A few remarks regarding alignment: The most obvious adjustment points are P2, P3 and C24. The potentiometers control the level of the unfiltered and filtered MPX signal respectively, whilst the trimmer varies the frequency of the FM oscillator. Preset PI influences the channel separ- Figure 3, Circuit diagram of the multiplex encoder. The 38 and 19 kHz signals are derived from the stereo decoder 1C, the MC 131 OP. Figure 4. The FM generator basically consists of an extremely stable oscillator in the shape of a single dual-gate MOSFET, the output of which can be frequency modulated. Figure 5. A simple but perfectly adequate power supply circuit. R43 and the LED, D9, are included to provide visible on/off indi- cation. ation. .Although this is always at least 40 dB, Pi can be used to compensate for negative crosstalk caused by the output filter L1/L2/CI3. SI should be closed and PI adjusted for maximum separation. If the unfiltered MPX signal is used, SI can be left open and PI will of course have no effect. Preset P4 allows the frequency of the oscillator in the stereo decoder 1C, MC1310P, to be accurately adjusted. The procedure is as follows: The output of an FM receiver tuned to a stereo transmission is used to provide the L and R input signals, whilst the MPX output of the stereo encoder is fed to an amplifier. P4 is then adjusted until a clearly audible beat note is produced as a result of the 1 9 kHz pilot tone in the L + R signal and the 19 kHz tone generated by the MC 1310P. Finally, we make no apologies for emphasising that, under no circum- stances should the oscillator output be connected to an antenna. Although the r.f. signal level does not exceed roughly 1 mW, this would nonetheless be suf- ficient to ensure a transmitter range of approx, 100 metres. One would there- fore not only incur the wrath of owners of FM receivers living in the immediate vicinity, but more importantly one would be breaking the law! H 1-30 — elektof January 1979 passive oscilloscope probe F passive oscilloscope probe probe scope input Many users of oscilloscopes fail to get the best from their instrument simply through not using a proper input probe. It is not an uncommon sight to see 'scopes being used with ordinary unscreened test leads, or with targe lengths of screened cable hung on the input terminals. The unscreened leads may, of course, pick up all kinds of interference signals, whereas a long screened lead greatly increases the effective input capacitance of the oscilloscope — thus attenuating high-frequency signals from high source impedances. The latter problem can be overcome by using a passive oscilloscope probe. The 10:1 probe described here can be constructed from standard parts. If an oscilloscope is to present a true 'picture' of an electric signal, the connection between the signal source (be, the circuit being tested) and the 'scope must fulfil a few basic require- ments, In the first place, excessive loading of the signal source must be avoided, since otherwise the amplitude and/or the waveshape of the signal may be modified. Secondly, having ensured that the signal is not modified at the source, it is also important to preserve the waveshape as accurately as possible when it is passed through the connec- tion to the 'scope. To sum it up briefly: if a signal in a circuit is to be displayed on a 'scope, it is important to ensure that both wave- shape and amplitude remain unaltered in the circuit and that the waveshape remains undistorted throughout the connection to the 'scope. It is not so important to preserve the amplitude of the signal, provided the ratio between input and output signal level is known. In order to reliably fulfil these require- ments, probe, connecting cable and 'scope must be considered as a whole. Generally speaking, however, commer- cially available probes are suitable for use with commercially available 'scopes — even if they are made by different manufacturers — since the input im- pedance of oscilloscopes is fairly well standardised. Commercial probes do have one disadvantage for amateur use: they are fairly expensive, , * The passive 10:1 probe described here is relatively cheap to build, and its performance is quite acceptable for amateur use. Depending on the care taken in the construction, reliable results can be obtained up to at least 500 kHz . The circuit of the probe is shown in figure 1 , The input impedance of the 'scope will normally be equivalent to a 1 M resistor (R2) in parallel with a 30 p capacitor (C3). The impedance of the connecting cable can be represented by a further capacitor (C2), By adding R] and Cl in the probe, a divide -by- 10 frequency compensated attenuator can be obtained. Since R2 is 1 M, the 10:1 t attenuation ratio implies that R1 must be 9 M. This value can be obtained by series-connection of a 6M8 and a 2M2 resistor. In order to obtain a flat frequency response, the capacitive voltage divider (Cl , C2 and C3) should have the same 10:1 division ratio. As stated earlier, the passive oscilloscope probe fllektor january 1979™ 1-31 value of C3 will normally be approxi- mately 30 pF. The value of C2 can be estimated: one metre of coaxial screened cable will normally have a capacitance of approximately 50 to 150 pF. For instance, ‘uniradio 70’ cable has a capacitance of 67 pF per metre; for 50 Q coax cable type RG58U the figure is 100 pF per metre. The total capacitance of C2 and C3 in parallel is therefore in the order of 80 . . , 180 pF for a 1 metre test lead. To obtain the desired 10:1 ratio, Cl must then be 9 . , , 20 pF; since the exact value is not known , a 20 p trimmer capacitor is used. The advantage of a 1 0: 1 attenuator probe will now be clear: the load on the Figure 1. For optimal results, probe, con- necting cable and 'scope should be dealt with as one complete unit. Bearing input Im- pedances and cable capacitance in mind, the complete circuit is equivalent to a frequency- compensated voltage divider. The probe proper consists only of Cl (20 p trimmer capacitor! and R1* the latter actually rep- resenting the series- connect ion of two resistors (6MB and 2M2b Figure 2. Mechanical layout of the home- made probe. Although constructed from odds and ends like piano wire and a mains fiex connector, the results can be quite good. circuit under test is drastically reduced. The input resistance has been increased from 1 M (R2) to 10 M (R1 + R2); the input capacitance has been similarly decreased. The latter is perhaps less obvious, but to give an example let us assume that C3 (the "scope input capacitance) is 30 pF; that C2 (the cable capacitance) is 100 p; and that the capacitance of the probe tip is 5 p. If C 1 and R1 are omitted, the total capaci- tance loading the circuit under test is then 135 pF, However, the correct value for Cl under these conditions is 130 — p“^14,5pF; series-connection with 02 and C3 effectively reduces this to approximately 13 pF, The load capaci- tance at the probe tip is therefore reduced to approximately 13 + 5 = 1 8 p! Construction The probe circuit can be housed in a standard mains fiex connector. This is ideal as it has a hole at each end, space for the components and cable clamps. To make room, the internal dividers which normally separate the cable terminals can be broken out using a pair of pliers. The components can be mounted on a piece of Veroboard, and placed inside a screen bent from a piece of copper foil, with a single hole in it and the side of the connector to adjust Cl, The coax lead is connected at one end and fixed firmly under the cable clamp. The input connection to the probe is made by means of a coaxial socket into which the probe tip plugs. This means that probe tips can easily be changed if damaged or if a different type of tip is required. Probe tips can be made by removing about 5 cm (2") of the inner insulation from a piece of low-loss UHF coaxial cable. In other words, the outer insu- lation, the screening braid and the core are all removed, leaving the inner insulation. This is then sleeved with a piece of brass tube (available from model shops) and a piece of piano wire is inserted down the centre. Instead of piano wire, any other type of stiff wire that is a good conductor and suf- flently resistant to oxidation can be used (e.g. beryllium-copper), the idea being to obtain a reliable and sturdy tip. The complete assembly is then mounted in a coaxial plug, the piano wire being soldered to the centre pin and the brass tube being gripped by the cable grip. Finally, the end of the probe tip is sealed with epoxy glue to prevent the ingress of dirt and moisture, and the brass tube is covered with silicone rubber, heat -shrink sleeving, or some other suitable insulation. Figure 2 gives an impression of the completed probe. To set up the probe a good, clean square wave with a fast rise time is fed into it and Cl is adjusted until there is no rounding off nor peaking of the squarewave’s leading and trailing edges. M Programmable sound generator A few months ago, a new 'Complex Sound Generator' introduced by Texas instruments was discussed under the heading 'Appllkator' (Elektor, September 1978). That (C could be used to produce a wide variety of 'complex sounds'; from trains and planes to gun-shots and space war. The desired sound effect was 'programmed' by means of wire links and resistor and capacitor values, A distant relative of that !C is a new chip introduced by General Instrument Micro- electronics (G1M): the AY-3-8910, This !C is a programmable sound generator: it can emit a broad range of complex sounds under microprocessor control. Although primarily intended for use in conjunction with General tnsrumem's own PIC 1600-series micro- processor system, it will also interface easily with several other 8-bit or 16-bit micro- processors (e.g. Texas Instruments IMS 1000 and Intel 8000-series). Sounds ranging from musical notes for musical instruments and complex sound-effects for electronic games and broadcasting to jarring warning signals for security applications and ultrasonic signals for remote control etc. are all easily generated. GIM are so impressed with the performance and versatility of the Programmable Sound Generator {or PSG for short) that they are offering automatic demonstrations over the telephone to ail callers. To hear a demon- stration of the device simply call the special number in London: 01 - 439 7052, As illustrated in figure 1, the sound generator is linked directly to the microprocessor chip — jn this particular example, GIM's own CP 1600 is shown. The link between the two chips consists of an 8-bit data/address bus {DAO . . , DA7), three Bus Control Signals {BC1 , BC2, BD1R) and the reference clock signal * 1. Communication between the con- troller and the AY -3-89 10 is based on the concept of memory mapped I/O. To a micro- processor such as the CPI 600, the sound generator chip looks like a block of memory, organized as 16 consecutive memory locations. The base address of this block of memory Is determined by the chip select lines (CS0* , . CS2 ) . In addition to the links to the microprocessor, the AY-3-8910 is provided with two 8-bit parallel bidirectional data ports that are TTL compatible. Each of these ports corresponds to a chip that can be used either as input or as output. Although these digital in- or outputs will be useful in many applications, the main purpose of this chip is to provide an analog (sound) output. In fact, the 1C has three of these outputs, each of which can be individually programmed to produce any desired frequency and/or noise signal. The possibilities for programming the desired envelope are rather more limited: the AY-3-8910 contains only one envelope generator, common to all three analog outputs. Partly for this reason, the three outputs will normally be summed as shown in figure 1 to obtain one total (complex! sound. The envelope generator can be programmed to give either a 'one-shot' or a continuously varying envelope. The various possibilities are illustrated in figure 2* As can be seen, the envelope shape is determined by four bits, labelled (rather loosely, as will become appar- ent) 'hold', 'alternate ', 'attack' and 'continue'. The dotted lines in several of the envelope plots indicate 'channel off' — no output signal, in other words. The effect of these four control bits can be described approximately as follows: Hold: if this bit is 6 the envelope can con- tinue to change freely; if the 'hold' bit is 1, the envelope is fixed at the end of its first period. Alternate: the amplitude of the envelope signal increases and decays during alternate periods. Attack: when this bit is 'high' the envelope will 'attack 1 , l.e. its amplitude wil rise more or less rapidly; when the bit is low 1 the envel- ope will decay* Continue: generally speaking, when this bit is 1, the analog output will not shut off after one cycle; when the bit is low, the output is only present during the first cycle of the envelope* It should be noted that the envelope gener- ator in the AY-3-8910 is not an analog circuit: its (.internal) output signal is a 4-bit digital code. The wave shapes shown in figure 2 are analog approximations of the actual staircase output. Programming for sound The final output signals at the analog outputs A, B and C are determined by the contents of 14 registers in the 1C. Each of these registers corresponds to a 'memory location' {as far as the controlling microprocessor is concerned); the address of each register is equal to the base address (set by CS0 . . . CS2) displaced by the register number. The two I/O registers have no direct influence on the analog out- puts. The function of the various registers is illustrated in Table 1 : the effect of the con- tents of each register is summed up briefly. As envelope control can be seen, the output frequency for each of the three outputs is determined by twelve bits; the eight least significant bits are stored in the 'fine tuning' registers RQ . . , R2, whereas the four most signif leant bits are stored in the 'coarse tuning' registers R4 . . . R6, The period time of the analog output is proportio- nal to the total 12-bit binary number, i.e. the frequency is inversely related. Similarly, the period time for the envelope generator {Te in figure 2} is determined by the 16-bit number stored in registers R3 and R7, Register R8 is called the 'enable' register. Each bit controls a unique function, as shown in Table 2. The bits are 'active low', in other words, they enable the corresponding func- tion when they are logic 0. For instance, an analog signal is present at output A when bit 0 is at logic 0; the output is turned off when this bit is T 1 . Bits 6 and 7 determine the func- tion of the digital I/O registers {R4 and R15, respectively); (S for input and 1 for output. The five bits in the next register, R9, deter- mine the clock frequency of the internal digital noise generator. Register R10 contains the four envelope- control bits, 'hold', 'alternate', 'attack', and 'continue'. The effect of these bits has already been explained — see figure 2, The last three registers that can directly influence the analog output signals are R 1 1 . . . R 1 3. The contents of these registers control the envelopes, as shown in Table 3, Under the heading Applikator, recently introduced components and novel applications are described. The data and circuits given are based on information received from the manufacturer and/or distributors concerned. Normally, they will not have been checked, built or tested by Elektor. applikator elektor January 1979 «* 1-33 Table 1. organization register bits function RO R 1 8 fO 8 fl R2 8 f2 R3 8 fE R4 4 PO R5 4 PI R6 4 P2 R7 description Fine tunes frequency on channel A, 8 bits are proportional to period Similar to fO but channel B. Similar to fO but channel C, Fine tunes envelope period. Coarse tunes channel A (high four bits). Coarse tunes channel B (high four bits). Coarse tunes channel C (high four bits). 8 PE Coarse tunes the envelope period* R8 8 enable Each bit controls a unique function (active low): see table 2. R9 5 N clock Varies the frequency of the clock for the noise generator, RIO 4 envelope control Each bit controls a function in the envelope generator: see figure 2. R1 1 6 envelope A R12 6 envelope B f Each of these registers controls its respective envelope as shown in table 3. R13 6 envelope C R14 8 I/O A RIB 8 l/OB With the control bits of RIO set for the output mode, data can be written to these ports from the CPU H and latched. With the control bits set for input, data can be read into the CPU. Each port is independently programmable. Table 2. Enable bits R8 BitO: Tone on channel A. Bit 1: Tone on channel B, Bit 2: Tone on channel C. Bit 3: Noise on channel A. Bit 4; Noise on channel B. Bit 5: Noise on channel C, Bit 6: Register R14 (I/O A) Bit 7: Register R15 (I/O B) > 'active low' } 0 for input, 1 for output L signal 1$ actually a 4-bit binary number. For this number to determine the output signal level, some kind of digital-to-analog conver- sion is useful, A suitable D/A converter is in- corporated in the 1C; since human hearing works logarithmically, as far as perception of sound level is concerned, logarithmic D/A conversion is used for the envelope level. The result is shown in figure 3: as the binary number changes from 15 through 0 and back up to 1 5 the envelope signal is varied in such a way that the audible level appears to vary smoothly over a 45dB range. Finally, the Bus Control Signals — generated by the controlling microprocessor — are de- coded within the AY-3-8910 to control all bus operations: writing into and reading out of the internal registers. Lift: General Instrument Microelectronics: Prelimi- nary Information on the AY -3 -8910 Programmable Sound Generator. Of the six bits stored in each of these registers, the first two are used as control bits. If these are both at logic J 1\ the output amplitude (of the corresponding output) is constant: it is proportional to the 4-bit binary number stored in the remainder of the register. If either or both of the control bits is at logic 0, the level of the corresponding analog output s determined by the output of the envelope generator — multiplied by a constant factor (xl , x'A or x14). As mentioned earlier, the resultant 'envelope' Table 3 Envelope A, 8 and C six bits in R11 . . , R13 amplitude remarks determined by envelope generator \ constant |00j * envelope xl 01 - -|' envelope 10 envelope x% 11 D3O2D1D0 * amplitude determined by bits D3 . . . DO Under the heading A pplikator, recently introduced components and novel applications are described. The data and circuits given are based on information received from the manufacturer and/or distributors concerned. Normally, they will not have been checked, built or tested by Elektor. 1-34 - elektor January 1979 computers and chess The game of chess has long been regarded as a symbol of man's intellectual prowess. Until recently, the prospect of a chess- playing computer defeating a master-strength human opponent seemed remote. A few months ago, however, in a much publicised match an International Chess Master, David Levy, actually lost a game to a program from North America. The story of the match is recounted by Mr. Levy himself elsewhere in this issue. The following article takes a look at the background to computers and chess: how they play, their weaknesses and strong points, and speculates on the chances of Karpov being the last flesh-and- blood World chess champion! Thirty years ago, with the electronic computer still in its infancy and illus- trating above all else the First Law of Thermodynamics (‘Work is Heat ), the game of chess attracted the interest of a number of researchers in the field of artificial intelligence. The first person to actually propose how a computer might be programmed to play chess was the English M a the m atici an C lau d e S hann on . In 1949 he presented a paper entitled ‘Programming a computer for playing chess’, which was significant, both for the fact that it was the first paper expressly on this subject, and more importantly since many of Shannon’s ideas are still employed in the strongest chess-playing programs of today. Shannon’s interest in chess program- ming stemmed from his belief that the game represented an ideal test of machine intelligence. Chess is clearly defined in terms of legal operations (moves of the pieces) and ultimate goal (checkmate), whilst being neither so simple as to be trivial nor too complex to be unsusceptible to analysis. Board, pieces and moves Shannon suggested that the machine represent the chess board by assigning a location in computer memory to each square of the board. Each piece is then designated as a numerical value: +1 for a white pawn, +2 for a white knight, +3 for a white bishop etc; -1 for a black pawn, -2 for a black knight, and so on. These numbers are stored in the memory location which represents the square occupied by the corresponding piece. An empty square is represented by storing a zero in the appropriate location. A number of more recent pro- grams also adopt this method, with the exception that a 10x12 board is used instead of 8x8, and that a unique number (such as 99) is stored in all the off-board locations, thereby allowing the program to detect the edge of the board. This is illustrated in figure 1, where the addresses for each square are given in the top left-hand corner and the contents (before the game starts) of the memory locations are also shown. The program generates legal moves simply by noting the mathematical relationship between the addresses of the different squares. For example, the addresses for each square may be assigned as shown in figure 1. Then, to calculate the possible legal moves of, say, a king standing on el (square 25) one adds the offsets +1 , +9, +10, +1 1 , _] — 9 } _iq and “11 to that address. The program then checks the contents of these new addresses to determine the legality of the move. If the location con- tains the number 99, the square is off the board and the move illegal. If the location contains a positive number, the square is already occupied by a white piece. If the contents of the location are negative, on the other hand, the king can legally move to that square cap- turing an enemy piece (always assuming that the piece is not defended). Finally, a location containing a zero also rep- resents a legal move assuming that the corresponding square is not attacked by an enemy piece. Calculating the legal moves for a sliding piece such as a bishop, is only slightly more complicated. With a white bishop situated on square XY (e.g. 54, where X = 5 and Y = 4), the program examines computers and chess elektor january 1979-“ 1-35 address [X+ l s Y+ l] (i.e. 65), checks to see whether the contents of that location are zero, and if so proceeds to examine address [X + 2, Y + 2] and so on (if [X + 1 „ Y + 1 ] turns out to con- tain a negative number , then the bishop can move to that square , capturing a piece, but obviously can move no further along that diagonal). The ma- chine repeats the above procedure for [X — 1 , Y — l],[X - 2,Y - 2j etc, then does the same for [X - 1, Y+ 1], X — 2 , Y + 2] etc,, and finally for X + 1, Y - l], [X+ 2, Y - 2] etc. In this way the program can generate legal moves along all four diagonals of the bishop . Similar operations can be performed to determine the legal moves of all the pieces, although one must bear in mind that certain moves have to be checked for the existence of pins (is the piece pinned against the king, for instance) and the procedure is complicated when considering castling and capturing en passant . A more 'logical' approach The above approach is still adopted by many modern programs, although an Figure 1. The computer can represent the chess board by assigning a location in mem- ory to each of the squares on the board. alternative method which is particularly suited for use with large computers has subsequently been developed. This utilises the fact that certain large com- puters operate with 64-bit words. If one bit is assigned to each square, only 1 2 64-bit words suffice to represent the position of ail the pieces on the board. For example, one word will provide information on the position of all white pawns on the board by setting a bit to *T for each pawn that resides on the corresponding square. If a square is empty the bit remains unset (‘0). A second word will represent the pos- ition of all the black pawns, a third the position of both white knights, and so on. In addition to the position of pieces, these ‘bit maps 5 or ‘bit boards 5 as they are called can be used to represent other information. For example, one 64-bit word might represent all the squares attacked by white's pieces, another ail squares which are a knight's move away from the black king, and so on. The real advantage of this alternative approach can be seen if one considers the instruction set of a modem com- puter, containing as it does a number of ‘Boolean Logic 5 operations. These can be used to combine considerable amounts of information stored in bit maps. For example, assume that we wish to know whether white has a knight's move that will fork black's king and queen. One simply fetches two bit maps of potential knight's moves from the black king and queen respectively, and a bit map of knight moves from their present squares. Since the square may not be occupied by a white piece, a map of all white pieces is inverted and then ANDed with the first three maps. If the result is non-zero, then a forking square exists. Finally, this map is ANDed with a map representing all squares attacked by black pieces to determine whether the forking square is defended. It can be seen that the above operation takes very few program steps. Looking for good moves Having enabled the program to generate legal moves, there comes the problem of selecting the good from the bad. This is where the difficulties start to accumu- late. The most obvious approach is to have the program examine all legal moves by white, followed by all legal replies by black, all legal counter-replies, and so on to a fixed depth. This procedure, which Shannon called the ‘type-A strategy 5 does however suffer from a number of serious draw- backs. The number of legal moves in each position is on average around 38. This means that a 2 -ply analysis of all legal moves (i.e., one move each side; ply = 1/2 move) would produce 38 2 - 1444 terminal positions to be evaluated. An analysis only 4-ply deep would yield 2,085,136 terminal positions, and a mere 6-ply look- ahead would involve evaluating some off board / Hio Tti3 "Jii4 Tub The “Tti? fiis ' [ 99 j 99 | 99 | 99 ] 99 99 j 99 j 99 , 99 tfoo”” +101 t|0f"7l03 [tM j"l05 ho6 |" 107” "has - *T fl? off board 14 15 i r i ^ i 3=7 i ™ ! 99 ! — i — i — i *1)6 ” “jbi I" da “ " *03 *04 |05 tbT “ TqY Toi" ” "!bi ~~ j 99 j 99 | 99 | 99 ! 99 ] 99 | 99 | 99 j 99 [ b f 79047 1 1-36 — elektor January 1979 computers and chess 3,0 1 0,936,389 positions! Because of this 'exponential explosion’ as it is called, an exhaustive look-ahead of this type rapidly becomes unmanageable, A second disadvantage of a fixed-depth exhaustive look-ahead is that the machine may well terminate its search in the middle of a series of exchanges - with the result that its evaluation of the position will be hopelessly wrong. It may, for instance, be deceived into thinking it is a piece ahead, when in actual fact it is about to loose a piece back — or even worse, A fascinating example of this type of 'computer blindness’ will be given later - in the game COKO v, GENIE. Shannon was well aware of the inherent problems of a type- A strategy, and therefore proposed an alternative model which he called type-B strategy. The latter is characterised by the notion of ‘quiescent 1 positions, i.e. the program is encouraged to continue its search until all forcing variations are exhausted and the position for evaluation is ‘static’, More importantly, a B-type strategy will not attempt to generate all legal moves in a given position, but rather will select a small number of ‘plausible’ moves for subsequent analysis. This approach obviously requires that the program have some criteria by which it can select the more promising moves from those which are plainly irrelevant, i.e. that the program have a ‘plausible-move gener- ator’. The interesting feature of the type-B strategy is that it attemps to simulate the approach of the most efficient chess model we know of, namely the human. Contrary to uninformed opinion, the chess master does not analyse dozens of moves deep and investigate hundreds of different variations before selecting a move. Quite the reverse is true. Research carried out by a Dutch psychologist, de Groat, revealed that, in a fairly typical middlegame position, Grand- masters tended to look at only three or four different possible moves, and that the maximum depth to which they calculated was not much more than 7-ply! However, the grandmaster is adept at perceiving the critical features of a position and at selecting an appro- priate plan. The grandmaster’s assess- ment of the position is liable to be much more nuanced than that of the amateur; he has ‘seen’ more deeply and recognised the truly salient, functionally important features. The story is told of the great Czech grandmaster Reti, who, when asked how many moves ahead he normally calculated during a game, replied ‘as a rule, only one*. Grand- masters think much more in terms of general strategy and the formulation of suitable plans than in terms of specific sequences of moves. For the chess programmer this knowl- edge comes as something of a blow, since pattern recognition is a task at which computers are as yet woefully inept when compared to humans. The difficulties in creating an effective pos- ition evaluator and plausible-move gener- ator are enormous, particularly when one bears In mind that the nature of chess is such that the failure to make one important move is often sufficient to lose the game, and clearly any move which fails an initial plausibility analysis by the program will never be played. However before considering more fully the problems posed by position evalu- ation, let us first examine how the com- puter actually goes about selecting a specific variation as the best available. Growing trees Shannon suggested that the program adopt the 'minima x 1 procedure first proposed by Morgenstern and von Neumann in their work on game theory. Basically, the program grows a ‘tree’ of variations. An example of a simplified game tree is given in figure 2, which starts from the initial position with white to move and assumes that some form of static evaluation function awards positive values to positions favourable to white and negative num- bers to positions favourable to black. The program assumes that, at each branching point {or ‘node’), the player who has the move will select the most promising alternative. That is to say that, when it is white to move (odd nodes), the program selects the variation leading to the largest evaluation, and with black to move (even nodes) it picks the branch which gives the smallest evaluation, The program first examines 1. e2-e4, e7-e5 2. Ngl-f3, Nb8-c6, evaluates the resulting position and stores the value thus obtained. Next it proceeds to evaluate Ke2-e4,e7-e5 2. Ngl-f3, d7-d6, and compares the result with that obtained for node 5. The lower of the two values is obviously best from black's point of view (remember that it is black’s move and the program is mini- mising at even nodes) and so that value is ‘backed up 1 to its immediate parent node (node 4) t The program then pro- ceeds to successively examine the two terminal positions (8 and 9) arising from node 7, evaluates them both, and backs the smallest of the two up to node 7. This procedure is repeated until the best ‘backed-up 1 values for nodes 11, 14, 19, 22, 26 and 29 are obtained. Next the program maximises at nodes 3, 10, 18 and 25 to find the best white move at that level. This process is continued, ‘minimaxing’ back up the tree, until the best move for the current position is determined. computers and chess etektor january 1979 — 1-37 Figure 2. A simplified game tree. Although this procedure seems ‘logical’, a full- width search to the depth shown here (4 -ply) would produce, on average, some two million terminal positions for evaluation. Fortunately, subsequent research showed that techniques can be employed which result in a substantial pruning of the game tree. A more funda- mental problem, however, is presented by the 'bottom line’ of the tree: in order to select good moves, the program must first evaluate the terminal pos- itions. Evaluating positions Shannon’s paper provided a simple example of an evaluation function which could be applied to static positions* Not surprisingly, the greatest weight was given to material balance and the rela- tive value of the pieces were assessed as 200, 9, 5, 3 and t for the king, queen, rook, bishop/ knight and pawn respect- ively* Positional evaluation was then incorporated by penalising doubled, backward or isolated pawns (- —Yi) and rewarding mobility by adding 1/10 for every legal move. Shannon also suggested additional features which should be included in the evaluation function, such as control of centre, open or semi-open files, passed pawns, pawn structure around the king, and so on. It is important that one arrive at an accurate weighting of the various factors in the evaluation function, and this is in fact one of the most difficult problems the chess programmer faces. Early programs in particular exhibited an alarming tendency to bring out their queens very early in the game, since this greatly increased their mobility score. However, as any beginner quickly learns, this is usually poor strategy . . , The problem of writing an efficient evaluation function is compounded by the fact that the importance of certain positional features changes during the course of the game. Furthermore, a particularly thorny problem is the difficulty in assessing whether a 'ter- minaT position is truly 4 quiescent'* or whether it in fact occurs, say, half way through a series of exchanges. As men- tioned, most programs attempt to resolve the latter problem by performing an additional search for all checks and captures until these are exhausted. How- ever, this approach is at best a makeshift solution, since it fails to deal with purely positional manoeuvres which a strong human player would examine as part of his evaluation of the position. In the position in figure 3, for example, the most significant feature is the ‘hole’ a b c d e f g h 79047 3 in Black's position at c6 to which White can manoevre his knight on f3. It is important that Black prevent this by playing Bh5 x f3. However, this is extremely difficult for a program to perceive. A further problem associated with evaluation functions is that many pro- grams contain an ‘opening book’ (i,e* lists of standard opening variations) which has been included by the pro- grammers to ensure a reasonable pos- ition from the opening. However, due to the un sophistication of the program's evaluation function, when the book runs out and the program has to start to think for itself, it naturally assesses the position quite differently from the Grandmaster whose game (or analysis) it is following* It therefore spends the next few moves re-arranging its pieces until they correspond with the evalu- ation function's idea of where they should be! Computers are greedy A typical fault of most programs is that they are excessively materialistic (even the Russian programs succumb to this capitalist evil) and are extremely loth to sacrifice a pawn or even a piece for less tangible, positional advantages. A start- ling exception to this rule occurred however in the first World Computer Chess Championships held in Stockholm in 1974. Favourite to win was a North American program called Chess 4.0, written by three former students of Northwestern University: Larry Atkin, Keith Gorlen and David Slate* In the second round Chess 4* 0, which hitherto had been undefeated by another program reached the following position (as black) against another North American entry, CHAOS : abcdefgh 79047 4 Black is a pawn ahead, having greedily consumed white’s king pawn, however he is behind in development, and in particular is not yet castled. UTiite seizes the opportunity to make a decisive piece sacrifice* What is surprising about this offer is that it must have been based on a purely positional evaluation of the resulting position, since the program could not possibly have seen sufficiently far ahead to ascertain that he would eventually more than recoup his invest- ment 16. Nd4 x e6! This move has been praised as ‘the finest ever played by a computer’ 17. Qe2 x e6 + {7 x e6 Bd6-e7 18. Rdl-el Qb8-d8 19. Be 1-/4 ■ ■#! ■ eat is Bf4-c7 19. m, * , Ke8-f8 20. Ral+dl Ra8-a 7 21. Rdl-cl N/6-g8 22. Rcl-dl a6*a5 Black has no good moves, white has a stranglehold on the position 23. Bf4-d6 Be? x d6 24. Qe6 x d6 + NgS-e? 25. Na4-c5 Bg6-f5 26. g3-g4 Qd8-e8 27. Bh3-a4 b4-b3 28. g4 x f5 And white eventually won, although it took another 47 moves to do so. Horizon effect v* snow- blind ness Selecting a suitable search depth also creates a number of problems when evaluating positions deep in the game tree. A particularly harrowing example of the fate that can befall a program when faced with a choice of equally promising continuations occurred in a now notorious — game between two programs called COKO and GENIE 1-38 — elektor January 1979 computers and chess which was played during the second ACM tournament in 1971. After the first 27 moves the following position was reached with COKO (white) to play: 790*7 5 COKO thought for 120 seconds and offered a pawn to entice the black king out into the centre: 28. c4~c5 + Kd6 x c5 ? Too greedy! COKO, having seen ahead the next 8V2 moves , played: 29. Qe4*d4 + ..... And announced mate in 8! Kc5-b5 Kb5-a5 Ka5-a4 ReS-dS + Rd8-d2 + Ra8-d8 + Rd8T3 = 8C557 ♦ 4 C3 n j i i 7| fC3 j i j pA7805 i Ll v. ! 12Qn 14 6 15 2 b 12 c 11 1 1 ir? ^ CA3161E * r 10 9 2 7 16 14 MSD o u NSD LSD a o :dpi DP2 o I I IM ASIC lO Ifg r Ijn ? DP3 C4 1?5n 73-005 7 Parts list. Capacitors: Semiconductors: Cl = 1 n T1 . , „ T3 = BC 177 r BC 557 or equ. Resistors: C2 = 270 n D1 ,02 = 1N4148 FH = 1 M C3 r C4 = 120 n IC1 - CA3162E R2,R3 - 1 k IC2 = CA 3161 E R4 . . . RG= 220 il R7,R8 = see table 5 Sundries: IC3 = M 7805 DPI . . . DP3 = see table 4 PI = 47 k cooling clip for IC3 P2 = 10 k SI - single-pole three-way Figure 7 . Complete circuit diagram of the universal digital meter. Figure 8. Printed circuit board end component layout for the complete unit {EPS 79005), 1-48 — elektor January 1979 universal digital meter used to enable the displays at the cor- rect moment in the multiplex cycle, via transistors T1 , . .T3. Virtually any common-anode seven -seg- ment LED display can be used. Several suitable types are listed in table 4* The ■decimal point 7 pin for each display is provided with a current -limiting resistor. Depending on the application, these can either be brought out to a selector switch or else one of them can be permanently connected to supply com- mon by means of a wire link. The basic measuring range of the circuit is -99 , . . 999 mV. By adding a voltage divider (R7 and R8), this range can be extended as required. Alternatively — and provided a voltage drop of up to 999 mV is permissible! — the 'universal meter 1 can also be used to measure cur- rent. In this case a suitable resistance value is chosen for R8, and R7 is re- placed by a wire link. The value for R8 is determined as follows: where If.s.d, is the desired full-scale cur- rent reading. For instance, if a 50 juA instrument is required, the correct value for RS would be 20 k. Table 5 gives values for R7 and R8 for several voltage and current ranges. It is advisable to use precision resistors (1% tolerance): the accuracy of the basic unit is 0.1% ±1 mV and the linearity is typically within 0,1 mV! There is room on the p.c. board to use two resistors in series for R7. If only one is required, the second position should be bridged with a wire link. Construction and calibration A suitable printed circuit board and component layout are given in figure 8. Any supply voltage from 7 to 15 V can be used ; the current consumption of the circuit is approximately 200 m A (all segments lit). If the unit is used as part of a larger system that already incorpor- ates a 5 V supply, the ‘on-card stabilis- ation" may be omitted: IC3 is replaced by a wire link between input and out- put As is apparent from figure 7 the input connection is floating, so that a symmetrical input is available if re- quired (e.g. if the unit is used in con- junction with the AC millivolt me ter described elsewhere in this issue). Note, however, that the maximum input volt- age range may not be exceeded! For most applications, the 4 0 ! input connec- tion should be connected to supply common by means of a wire link (shown dotted on the p,c. board). Calibration is, of course, important. For best results, some suitably accurate reference is required — an accurately calibrated digital meter or an accurately specified calibration voltage. Finding a suitable reference voltage source is not as easy as one might think. ‘Reference zener diodes" are not ac- curate: the normal tolerance is 5%, Specially-designed reference voltage sources, such as the National LH 0070 and some Analog Devices devices (sorry, we couldn’t resist that one), are rather expensive for this application. There are, however, two readily-avail- able alternatives. A miniature mercury battery, as used in cameras, hearing aids, digital watches and the like produces L37 V, within 3%. Using a voltage divider, consisting of a 4k7 and a 1 0 k resistor, a reasonably accurate reference voltage can be ontained; 0.93 V, ± 5%, Good enough for most purposes. Alternatively, the calibration procedure can be carried out using a multimeter as a reference. In both cases, however, the Least Significant Digit has no meaning and can best be omitted. Once the problem of obtaining a refer- ence meter or a reference voltage source has been solved, the calibration pro- cedure is easy: — short the input to ground (a wire link across R8); — adjust Pi until the displays read ‘ 000 ’; — remove the short at the input, and connect the reference voltage source; — adjust P2 until the correct display is obtained. lighting unit automatic emergency This unit charges a nickel- cadmium battery from the mains to provide a standby power supply for emergency lighting in the event of a mains failure. When the mains supply drops out, the lighting is switched on automatically* The circuit of the unit is extremely simple. Trl, D1 and 01 provide a half- wave rectified and smoothed DC supply of approx. 6 V, which is used to con- tinuously charge the Ni-Cad battery at about 100 m A via Ri and D2, A 2 Ah Ni-Cad can safely be charged at this rate. The voltage drop across D2 reverse- biases the base-emitter junction of Tl, so that this transistor is turned off and the lamps are not lit. When the mains supply fails, however, Tl is supplied with base current via R2; the transistor therefore turns on and the lamps are lit. As soon as the mains supply is restored, Tl will turn off, the lamps are extinguished and the battery is once more charged via Rl and D2. The unit can be mounted wherever the event of a power failure. An obvious example is in that infamous dark cupboard under the stairs, so that, should a fuse blow, a replacement can be easily found and fitted. A transformer with a slightly higher secondary voltage can be used, provided that Rl is uprated to limit the current emergency lighting will be required in through this resistor to 1 00 raA. Number and colour coded BCD OIL switches Erg BCD dual in-line switches offer an ideal means of hardware BCD programming. Of low profile body height only 7.9 mm maximum) each switch status is clearly seen at a glance by the position of colour coded arrow' heads on the moving switch elements that stand out against the white bodies. Each individual switch is numbered in a standard 1, 2, 4, 8 coding. The switches may be set up with the aid of a small probe, such as the blade of a small screwdriver and, since the switching members have a positive detent action, they cannot be accidentally moved. Capable of sw itching 1 jjV to 100 V, 1 mA to 1 A up to 10 VA, Erg BCD dual in-line switches have an initial contact resistance of only 8 mn (typical). A contact resistance repeatability of ± 1 mfi measured at only 10 mV/ 10 mA over 100 operations) is usual. Tested to BS2Q11 Ea for vibration, the switches will operate reliably over a temperature range of —55 3 C to +1Q0*C. Free samples of the Erg BCD dual in-line switches are available on request. Erg Industrial Corporation Limited '* Luton Road, Dunstable, Bedfordshire LU5 4LJ (978 Ml Programmable TV games General Instrument Microelectronics have introduced i new set of MOS microcircuits for use in cartridge-based programmable TV games systems. Known as SYSTEM 8601, the circuits include a clock generator, colour encoder, modulator and a election of cartridge micro- circuits - enabling fully programmable games systems to be built at low cost. Each games system will consist of i console into which individual gjjnt set cartridges are slotted. Each console will contain clock, encoder and modulator, as well as tame controls, switches, power supplies, etc. Each cartridge contains individual games micro- circuits, plus interface circuitry, *nd all sets will feature realistic sound generation and on-screen scoring. Some of the cartridge-mounted microcircuits are already available, including the 8610 Supersport’ (20 games), the 8765 'Motorcycle 5 (8 games) the 8603 Road Race 5 (3 games), the 8607 'Target’ (12 games), the 8606 'Wipeout’ (24 games) and 8605 'Warfare’ (10 games) - with more to follow before the end of the year. David Letheren, GIM’s Consumer Marketing Manager, believes that the SYSTEM 8601 will prove extremely popular with OEM’s and amateurs alike, the launch being well timed for the 1978 Christmas TV games buying peak. He comments: 4 The chief advantage of the low cost programmable system is that it offers the consumer a large number of game combinations with cartridge flexibility at a console price of approximately one-half that of the pro- grammables currently in the marketplace. We also intend to consolidate our market lead by adding between four and six additional cartridges to the system over the next twelve months.’ In 1976 G1M introduced the AY -3-8500, which has become the dedicated video game industry' standard. Since then the company has dominated the market by a combination of aggressive world- wide marketing and continual innovation. The development of the 8601 system came after extensive consumer preference studies. With several of the major games manufacturers already introducing models using this system GIM’s marketing plans appear to have been justified. General Instrument Microelectronics Ltd., Regency House, 1-4 Warwick Street, London W1R 5WB r England (979 m DC motor speed control Fairchild’s proprietary juA 7392 14-pin DIP DC motor speed control circuit is designed to provide precision, closed-loop, speed control of motors in systems where a tachometer reference signal is available as an indication of speed. It is ideal for design situations where current requirements are either less than 300 mA or greater than 2 A (w r hen drive can be provided through an external power transistor or power darling ton). The tachometer frequency can be generated in any manner, the only constraint being that the signal available at the device’s input terminals exceeds 100 mVp.p, Possible types of tachogenerator that can be used include; a multiple pole motor winding, an optical pick-up from the motor’s shaft or an optical or magnetic pick-up from a tape recorder capstan or record turntable. The juA 7392, on receipt of the tachogenerator signal, first converts it into a pulse with a defined width and amplitude, the pulse is then integrated to generate a sawtooth waveform. This sawtooth is then compared with a DC reference and a pulse width modulated signal generated the duty cycle of which is related to the error signal Average output current available from the M 7392 is up to 300 mA. The motor inductance provides adequate smoothing so ensuring what is, essentially, a direct current through the motor, provided the tachometer frequency is a sufficiently large multiple of the motor speed. Two distinct design advantages are offered by the juA 7392 system. Firstly, speed regulation is independent of the amplitude of the tachometer signal - it depends only on the frequency. Secondly, at higher battery voltages the system is more efficient than the equivalent DC control system. This results in extended life for battery powered equipment. Specific design features include precision performance. (freque n c y-to-vo 1 tage con vers io n stability is typically 0.1% for V+ from 10 V to 16V); on-chip thermal shutdown, over-voltage protection and 4 stailtimer’ facility ; wide supply voltage range 6,3 V to 16 V and a clamping diode available on a separate pin. Fairchild Camera <£ Instrument ( UK ) Ltd., 230 High Street, Potters Bar , Herts, EN6 5BU (976 M) MNOS non-volatile quad decade counter A new addition to the ‘NovoT (non-volatile MNOS) range of standard logic circuits is announced by Plcssey Semiconductors. This is the MN9 1 05 which comprises a 4-decade BCD up/down counter with the output of each decade addressed from two input pins and available on a common 4-bit data highway. In parallel with an MOS counter is a 16-bit MNOS memory into which the contents of the counter can be written by applying a SAVE signal to the circuit. When this data has been written into the memory it is retained even in the absence of applied power, and then it can subsequently be recalled from the memory to preset the counter. Data storage in the absence of applied power is guaranteed for at least one year at temperatures up to 7CFC, and the guaranteed number of write operations is 10 . The device has a count frequency range from DC to 250 kHz. A major feature of the MN9105 is that it requires only normal MOS operating supplies of +5 volts and -12 volts. All inputs and outputs arc TTL/CMOS compatible. The MN9105 has a wide range of applications, for example as replacements lor electro- mechanical batch counters in vending machines, or time elapse indicators. PJessey Semiconductors, Cheney Manor, Swindon, Wiltshire, SN2 2Q W (985 M] UK 16 - elektor january 1979 market mvrrnrw? 9 Watt audio amplifier A proprietary Fairchild design the mA 783 audio power amplifier is designed for high voltage (24 V) applications driving 8 and 16 ohm loads. Encapsulated in the standard 12-pin package two heatsink configurations are available. Designed for use as a low frequency class B power amplifier it can provide 9 W into an 8 ohm load (typically). i ..... „ wM 1 The fiA 783 is able to operate from a wide supply voltage range , with a maximum of 30 V, A high output current (repetitive) capability of 2*5 A is also exhibited by the device* An on-chip thermal limiting circuit offers the designer the following two advantages; 1) an overload on the output, even if permanent, or an above-limit ambient temperature can be easily handled, 2) the heatsink used can have a smaller factor of safety compared with that of a conven- tional circuit. Should the junction temperature rise too high, power output, power dissipation and the supply current decrease so protecting the device. Typical applications for the 783 include tv audio circuits, inexpensive radio receivers etc. A design point worth noting is that by operating audio amplifiers at higher voltages simplifies power supply filtering problems. Fairchild Camera & Instrument (UK) Ltd , 230 High Street , Potters Bar , Herts EN6 5BU 1925 Ml High speed 128 k word core memory Dataproducts (Dublin) Limited announce the availability of a new 128 k, 18 bit word core memory module. To be known as Maxi-store, the new memory module has been designed to combine high speed access time with mass storage - a 325 nanosecond access time and 750 nanoseconds cycle time. Cost per bit with Maxi-store is in the 0.2 cent per bit range for OEM quantities. The Maxi-store modules are designed for mini and midi computer applications where high speed large storage is a principal requirement. The Maxi- store is expandable to 1024 k words. Using a 3 wire, 3D organisation, the basic Maxi-store module is on a self sufficient planar card, with the address, data registers timing and control logic. The module operates in the read/restore, clear/write and read/modify/write modes. It requires only two voltages: +5 volts DC and + 15 volts DC. Up to 4 Maxi-store modules can be housed in a 1 OVa" (26.7 cm) high by 19" (48.3 cm) wide by 24" (61 cm) deep rack mountable chassis. The chassis also contains a power supply and an extra slot for either a self-test card or a custom interface card. A memory protect circuit provides data saving upon loss of power. Optional chassis wiring permits 36 bit double word configurations or daisy-chaining tw o chassis for expansion to 1024k by 18 bit capacity. The Maxi-store may be purchased as a complete system or as an individual module. Dataproducts (Dublin) Limited, Telephone (01) 31166 Dataproducts International Inc., Tel: Reading (0734) 58723 - 6 (931 M) Handheld 3-1/2 digit DMM controlled by CMOS microcomputer The first handheld 3-1/2 digit multimeter to be fully controlled by a CMOS microcomputer chip has been introduced for less than $ 400 by Electro Scientific Industries of Portland, Oregon, Called the Calcumeter 4100, it is essentially a high performance calculator integrated with the multimeter to enable some extraordinary capabilities: 1) Save the average design engineer hours per week by providing the ability to automatically and directly scale and offset (mx + b); sort with high-low limits; average noise away; measure dB volts directly; display in percent deviation; troubleshoot by sound. 2) Measure and HOLD one million times on a single 9 V battery. 3) Control and datalog remotely with an accessory printer. 4) Perform math conversions with 11 special keys. 5) Store measurements and/or calculations in five different memory locations. 6) Operate with autoranging, autozeroing and autopolarity. 7) Select from three different display formats: Scientific notation, engineering notation (exponents in multiples of three) or any fixed decimal up to 7. The multimeter operates in six ranges as follows: 10 microvolts sensitivity through 1000 volts DC, 750 AC; 10 microamps sensitivity up to 200 milliamps (extended to 20 amps with accessory shunt); 0, i ohms resolution through 20 mega- ohms, Basic accuracy is 0,25% DC V. *• ^ 5 ■«»•§ »f|| j III :t| : -l!! fi|- fll. |||: |!§ : | m iil" l!i ss': ■:$ p 4pf ^*;|||.||| frip:; If -If | . § imsn 8-bit bi-directional bus-buffer A new 8-bit TRI-STATE™ bus transceiver from National Semiconductor provides bi-directional drive for bus-oriented microprocessor systems. Offered in a single 20-pin DIP, the 1NS8208B device is manufactured by low power Schottky technology. The 1NS8208B is part of National's expanding Series 8000 family of microprocessor peripheral digital I/O, peripheral control, communications and memory support products. All products are compatible with National's Universal MicrobusTM concept as w r ell as all other bus-oriented microprocessor systems. The chip has 48 miliamperes drive capability on the B-port (Bus- transceiver) and 16 mA drive capability on the A-port; an additional PNP transistor input on both ports allow s reduced input loading. Typical short-circuit output current is 38 mA for the A-port and 50 mA for the B-port. For 300 pf load, the A to B-port propogation delay is 18 nano- seconds for logical '0% and 16 nanoseconds for logical T’ transition. Each receiver section requires a minimum of 2 volts at only 0.1 micro -amperes (typical) for a logical T* signal; logical 'O' requires 70 mA. The INS8208B has lower supply requirements than most other available bus transceivers. The pow T er supply requirement will not exceed 130 mA. For simplified system inter- connections, the 1NS8208B Available since September, the standard product comes with test leads (with fingerguard probes and recessed connectors), direct prod for probing with instrument in hand, two alligator clips that screw r on to the end of the probes, a complete Owner’s Handbook consisting of more than 100 pages, shortform manual which snaps into a plastic storage case (the latter serves as a bench top cradle), and a spare fuse. Micrometries , Inc., of Portland, Oregon; Suntek Business Park , 9450 S.W, Barnes Rd., Portland, Oregon 97225 1989 M) transmitter and receivers, connected as reversed pairs in parallel, are applied to tw o sets of eight I/O ports. Only two control signals arc required - a transmit/ receive signal to enable the trans- ceivers and a chip disable signal which places both sets of ports in a TRI-STATE condition. National Semiconductor GmbH, Industrie strafe 10, D-8080 Furstenfeldbruck, West Germany (977 M) market eJektor January 1979 — UK 17 fyf;TiT!TWT W \ WWVWI mTw& A Pocket terminal The GR Electronics Pocket Terminal is the most practical low cost, hand-held communications unit yet developed, with a wide range of input/output facilities and multiple signalling options. Its total portability and rugged packaging will make it an essential item of equipment for tech- nicians, programmers and operators - in fact anyone with a need to communicate with computer systems or stand-alone processors. Features * 1 6 -segment ‘staiburst" LED dis- play providing instantly legible 64 -character ASCII alphanu- merics and symbols * All 128 ASCII codes generated from positive action keyboard * 30-character memory dis- playable In eight-character blocks * Single 5 V supply required at 350 mA, typical * 110 or 300 baud transmission rates, internally selectable * Full duplex operation, with interface for 20 mA loop or V24/RS232 levels * Parity codes and stop bits settable to suit your trans- mission standard * Internal ‘bleeper’ reacts to ‘BEL 1 code; unit also responds to cursor controls, ‘display clear" and ‘new line’ codes Applications As well as its use in conventional small-scale I/O operations, the Pocket Terminal opens up a new range of applications made poss- ible by its portability, low cost, convenience and flexibility, * In situ fault diagnosis on processor-based systems * Clear, unambiguous mobile communications * On-site reprogramming of operating parameters * Comm is sio ning o f d igital systems * Interactive debug, information retrieval, status monitoring, etc. * Bench testing • Educational and home computing Description The unit is a hand-held terminal with a 40-key positive tactile response keyboard comprising two single- function and 38 dual- function keys. These give internal control and allow transmission of all 1 28 ASCII codes, with a maxi- mum rate repeat facility. Audible response to an external signal is also provided by an internal ‘bleeper’. The display is of the 16-segment ‘starburst" type, with capacity for eight characters in line. Alpha- numerics and symbols arc conven- tionally formed, and the full 64-character upper case ASCII set may be generated. As characters are received via the data link they are stored in the unit’s memory, which may be visualised as a 32 character line. The display acts as an eight character ‘window" onto the 'memory line*. Far left and far right positions in the line are reserved for display of the internal control status, i.e, ‘shift", ‘control’, ‘repeat 1 , leaving 30 pos- itions for data received* The display window may be stepped left ox right in blocks of four characters along the line, or in a single move to the left home’ or ‘right home" limits. Location of the display window in the line is indicated on the display. The terminal allows two modes of operation* In the first, each character received is entered in the memory at a position deter- mined by a cursor which may be controlled remotely by received codes. This allows data to be entered in any required format, as w ell as permitting data already in the memory to be edited. In the second mode the cursor is static and information is input at the right hand end of the memory line, ‘shuffling" existing data in the line one step left as each character is received* If the line is filled in the first mode, the terminal will automatically enter the second mode. A removable panel on the rear of the unit gives access to a switch set allowing the following options to be selected : 1 * Single or dual stop bits 2. Control code response enable/ disable 3. Parity bit EVEM /ODD/SET/ RESET 4. 300/110 baud transmission rate OR Electronics Ltd Fairoak House, Church Road , Newport, Gwent NPT 7 EJ Motor protection relay A range of temperature-sensitive motor-protection relays which have a rapid response- time with negligible overshoot, has been announced by P & B Engineering Co. Ltd., of Crawley, Sussex, England. Called the Model MW Golds’ Relays, they protect industrial multi-phase motors against damage from overloads or failure of a phase in the mains supply. The relays also provide a continuous indication of the percentage of full-load current at which the motor is operating. Additional relays which give protection against earth-leakage and short-circuit damage can be fitted. The new thermal relays are be- tween 25% and 50% less expens- ive than solid-state electronic motor-protection devices and are extremely reliable* Fast operation is achieved through the use of a special contact-mechanism which is actuated by three bi-metallic coil assemblies heated directly by current derived from transformers in the motor’s supply. The coil assemblies have a heating curve similar to that of most in- dustrial motors. Each assembly comprises two bi-metallic coils, one mechanically-linked to a contact of the contact mechanism, the other providing temperature- compensation. The coil of the centre phase is linked to the ‘H ’-shaped centre frame which carries the outer pairs of tw r o sets of contacts. The centre contact of each set is linked to one of the other two phases. Under normal operating con- ditions, all three coils deflect through the same angle, so that the centre contacts of each set remain mid-way between the outer contacts. Should the three phases become unbalanced — through the failure of a phase, for example the bi-metallic coil of the affected phase slowly reverts to its zero position (i,e. goes cold). At the same time the increased currents through the other two phases deflect their coils further, so that in a matter of seconds, the contact of the affected phase touches that of the adjacent phase and completes the trip circuit. Tripping also occurs when the currents in the outer and centre phases differ by approximately 1 2% of the full load value. The MW relays have a power consumption of only 2VA per phase at full-load. Their contact mechanism can be set to trip at any percentage of full-load current between 80% and 125%. Accuracy of setting is ± 3%; repeatability at a given trip-setting is better than + 1.0%. Contact assemblies can be fitted with auxiliary contacts for actuating alarm systems and other switching fuctions. The relays are housed in dust-proof cases and can also be supplied in withdrawable cases for ease of inspection and mainten- ance. P & B Engineering Co. Ltd. is looking for agents and distribu- tors in France, Italy and Germany for its range of products which, in addition to thermal motor protect tion devices, includes maximum- demand load indicators as well as portable earthing equipment for earthing electricity sub-stations and overhead-lines up to 400 kV during maintenance. P&B ENGINEERING CO r LTD . 29/31 Kelvin Way, Crawley , West Sussex RHI0 2PT t England (1026 M) Diode for fiber optics The new infrared transmitter diode FV 21 IR from Siemens with flat epoxy encapsulation of the light-emitting GaAs chip has been designed specially for optical-fiber cables. The glass fiber ends can be attached directly, the wavelength of the IR beam is 880 nm. The chip is mounted on a 1046 base plate, the cathode being metallically connected to the housing. l he diode provides an optical power of 1 to 2 mW at a lDQ-mA drive. The housing has a diameter of 5.4 mm and a height of 1.3 mm. Siemens A G, Zentrahtelle fur Information Postfach 103 D'8000 Miinchen 1 Federal Republic of Germany (1037 Ml 1983 Ml lOp lOp lOp 12p 12p 25p 25p 12p 12p 12p 15p 15p 25p 45p 25p 25p 12p 20p 15p 20p 20p 22p 22p 25p 12p 20p 28p 20p 20p 12p 45p 40p GOp 60p 65p 50p 50p 50p 12p 12p 12p 12p 12p 26p 20p 2Sp 25p 26p 25p 40p 85p 76p 75p 70p 6Qp 2Sp. 130p 25p 40p 3Sp 30p 70p 46p 46p 120p 74100 80p 74104 40p 74105 40p 74107 25p 74108 lOOp 74166 75p 7401 740? 7403 7404 7405 7406 7407 7408 7409 7410 7411 7412 7413 7414 7416 7417 7420 7421 7422 7423 7425 7426 7427 7428 7430 7432 7433 7437 7438 7440 7441 7442 7443 7444 7445 7446 744 7 7448 7450 7451 7453 7454 7460 7470 7472 7473 7474 7476 7476 7480 7481 7482 7483 7484 7485 7486 7489 7490 7491 7492 7493 7494 7495 7496 7497 121 122 123 125 126 128 130 131 132 135 136 137 138 141 142 143 144 145 147 148 150 151 153 154 156 156 157 160 161 162 163 164 165 166 167 170 173 174 175 176 177 178 179 180 181 182 184 185 188 190 191 192 193 194 195 196 197 198 199 293 LSOO SI 12 7805 7817 7815 7818 7874 lOOp lOOp lOOp LINEAR 4000 12p 4040 60p 4001 12p 4043 60p 4002 12p 4046 90p 4006 80p 4047 SOp 4007 14p 4048 SOp 4009 30p 4049 2Sp 4011 12p 4050 2Sp 4012 I2p 4054 lOOp 4013 30p 4055 I30p 4015 5 Op 4056 1 20p 4016 30p 4060 lOOp 4017 50p 4066 35p 4018 55p 4069 12p 4019 4 Op 4070 12p 4020 50p 4071 12p 4022 50p 407? 12p 4023 12p 4081 12p 4024 40p 4082 12p 4025 12p 4093 70p 4026 80p 4510 60p 4027 30p 4511 70p 4028 45p 4516 65p 4029 50p 4518 65p 4030 30p 4520 65p 4032 80p 4528 80p 4033 lOOp 4583 70p CLEARANCE OFFERS Assorted Japanese IF Transformers 20 for £1.25 Assorted Ceram< Capacitors (no rubbish) 300 for £1.30 Hook-up Wire in solid black 100 yards for £2.00 Assorted transistors BC147 8C1S7 BF194 BC148 BC158 BF 195 BC149 8C159 8F196 BF197 100 for £6.00 Reed Inserts 28 mm normally open gold contacts. 10 for £1.00 100 lor £7.00 POWER SUPPLY CAPS 770C lb 7700 40 2700 63 7700 100 3300 30 3300 63 4 700 40 4700 26 50p 1 20p 4700 63 80s. 4 700 70 150p 10000 '0 lOOp 10000 75 ISOp ’5000 16 150p 60p Wp 77000 25 TOOp 60p ENOU'RtS FOR A*v ()Tm*H TvPfS ELEC CAPACITORS AY 38500 450* NE556 90p C A 3039 70p NE562B 400* CA3046 SOp SAD 1024 1500* CA3060 SL917B 650* CA3065 200p SN76003N 150* CA3076 250p SN76013N 110* CA3080 75* SN76013NO 125* CA3084 250p SN 76023 N 110* CA3085 85* SN76023N0 CA3066 60p SN76033N CA3088 190* SN76227N 160p CA3089 1 60p SN76228N 180* CA3090AO 360* SN766EON 75* CA3123E 130* TAA300 100* CA3130 100* TAA350 190* CA3140 60p TAA550 35* LF356 SOp TAA570 220* LF357 80* TAA661B 140* »Op T A A 700 350* LM3002R5 170p T A A 790 360* LM301AN ' JUp TA0100 150* LM304 200p TA0110 130* LM307N 65* TBA120S 60* LM308TO5 lOOp TBA120T 86* LM308DIL 100* TBA480O 200* LM309A 100* TBA520Q 200* LM310TO5 150p TBA530Q 200* LM311T05 150p TBA540 200* 325* TBA550Q 250* 70* TBA560C 250* 60p T8A641A12 250* LM348N 90* TBA700 180* 60p T0A72OQ 225* »p TBA750Q 200* 90p TBA800 80* iaop TBA810 1CX1* 25* TBA820 100* 40* T8A920Q 280* LM710T05 60* TCA270Q 220* LM710DIL 66* TCA270S 220* LM723T05 40p TCA760 300p LM723DIL 40* TCA4500A 460p LM733 120* TDA1008 J50p LM741 20* TDA1034 460p LM748 40p TDA2002 300p LM1303N lOOp TDA2020 300* LM 1 458 100* TL084 120* LM3080 75* XR320 250* LM3900 56* XR2206 460* LM3909N 86* XR2207 460* MC1310P 140* XR2208 600* MC ’ 3 1 2P 150* XR2216 650* MC1314P 190* XR2567 250* MC1315P 730* XR4136 150* MK50398 660* XR4202 150* MV5314 380* XR4212 ISO* MM5316 480* XR4 739 150* NE529K 150* ZN414 100* NE555 26* 95H90 700p SPECIAL OFFERS 100 off 741 100 off 566 s 100 off IN4148 100 off AD161 100 off A0162 100 of .125 Red Leds 100 off .2 Red Lads IfiOOp 19QOp ISOp 2500p 2500p 750p 750p 100 off Grpen/Yeilow Led* 1 200p 1 50 2 7 75 2 7 35 3 3 75 4 7 10 4 7 16 4 7 ?5 4 7 50 6 8 75 10 10 10 16 10 75 10 50 72 6V 3 77 10 7? 16 7? 75 7? 35 7? 50 33 6v3 33 16 33 25 33 40 33'50 7p 7p 7p 7p 7p 7p 7p 7p 7p 7p 7p 7p 7p 7p 47 75 4/ 35 47 50 100 10 100 16 100 75 100 60 100 63 770 16 770 75 720 50 330 25 330 35 330 50 470 10 470 75 470 35 470 50 1000 16 1000 25 1000 35 1000 40 1000 63 1 700 63 7700 10 27p 1>P 1«P 20p 2*9 77p POLY CAPS 1000 PF 7200 3300 4700 6800 0 01 uf 0 077 uf 0 033 uf 3047 uf 0 1 uf 0 22 uf 0 33 uF 047 uf 1 0 of 2 2 uf 4 7 U F 6 8 uF IOuF 7* •p 12p TANT. BEADS 01 /36 V 0 15/36V 0 27/36V 0 33/36 V 0 47/ 10V 0.47/36V 0 66/36 V 1 00/ 10V i 0Q/36V 1 5/36V 2.2/26V 2 2/36 V 3.3/16V 4 7/16V 4. 7/26V 4 7 /36V 68/6V3 6 8/35V 1 0/36V 22/ 15V 33/16 47/3V 47/16V 100/3V ASSORTEO GIANT SCREW PACKS INCLUDES SELF TAPPING. SELF CUTTING SCREWS. NUTS. BOLTS. WASHERS. EYELETS ETC. ETC. WEIGHT 1kg 2.2lbs. Approx 1400 items ONLY £1.80 (U.K. only) At. 17/ Ol At 128 AIH6I At) 167 A0 16 • / as i m AS 73« Af 27V At ll 10 MAI 14 HA 1 7 • MA 164 MAI67 HAI / f MAXI | HBI06 hAi m 8007 8008 aooec Bt. '09 8O09C BO»3 8014 BCH6 8016 BO 1 7 BO’B BOI9 80 75 BC 1 258 BO 76 80 34 80 36 8037 80 38 BC 140 BC141 8047 BO 43 BC 14 7 8048 6O40C 8049 BC 153 8054 8057 8058 8059 BC 16/A BO 68 BC 169 BO 71 1* 12p 12p lip 1 2p 17p 15p 25p 76p 75* Itp 75p 65p 50p 60p Ml 176 HF 17/ HF 1 (/ HI I 7R H» I 79 HI IHfl Ml 1 HI M» I H7 Ml 'HI Ml >84 HF iMh HI 194 Ml 196 Pf 196 H» 197 M* 774 HI 776 Ml 741 Ml 744M Hf 766 Ml 76 7 Ml 76M M» 759 M» 774 HI 174 HF 116 Ml 11/ Hi 36/ Ml 394 BF451 HI 4 68 H* 694 Ml 696 8169/ BF R40 B* 880 B» W58 0FW6O 81M88 BF W89 BF 8f9i BF *79 BF *34 BF x J8 81 X44 Ml X4H BF *81 BF *86 Ml *87 BF X 88 HF y 10 Ml y 18 81 >50 BF *61 81 »6? 8F y63 81 y 90 BRIO! 8RyJ9 B Fr * 66 BS*95 BT100A BU* 06 BUH3 BU70B 8*»00 I 110 I 470 E430 V if 3411 VH6A06 MPSAOfo MPSA56 UP79 I IP79A 1lH?9t • IPX) TlPTOA IH'IOH llPTOt. UPJ1 IIPJIA 1IPJ1H FiPTIt tlP17 t IP4 1 H riP4it f IP47A MP47H f IH47* MP7966 I IP1056 f 1690 1 169 1 IN914 IN3/64 •N4001 INA007 IN4003 IN4004 IN4006 IN 4006 iM»00> IN4148 7N466A 7N979 7N930 . 7N1307 7N1303 7N1305 7N1306 7N1308 7N17M 7N7719A 7N7483 7N7906 7N7907 7N3051 7N3054 7 N 306 6 7N3702 7N3703 7N3704 7N3705 7N3706 7N370B 7N3716 7N3819 7N3866 7N3904 7N607 7 7S0734 7N6777 MURATA ULTRASONIC TRANSDUCERS 40k Mi Type MA40 LIS Transmit Type MA40 LIR Receive £2.00 each £3.50 pair Data lOp ROTARY SWITCHESBY LORLIN 1 POLE 12 WAV ? POLE 6 WAY 3 POL f 4 WAY 4 POLE 3 WAY All at 40p Each 5p 6p 5p 6© Jp 4p «0p 20p 70* 75* 50p ?2p ?Sp 30p 16p 70* ?0p 60p 12p i2p 15* 5Qp 50* OPTO ELECTRONIC CORNER SPECIAL SCOOP Of FER 0 »?S or 0 2 -nch RED LEDS IS* nch. 10 for £1 00. 100 tor £7.50, 1000 for £60.00 0.125 or 0.2" Yellow* and Green LEDS 16* lOfor £1.40, 100 for £12.00 HP5082 - 7750 160p each DL747 Seven Segment Common Anode Displays Character Height 0.6" £2.00 eech FND500 Seven Segment Common Cathode Displays Character Height 0,5" £1.30 each 4 tor £5.00 2N5777 Photo Carhngion CONSONANT 9945 PR E CONSONANT LUMINANT 9949 Complete set of parts including P.C.B Transformer and front pane' Cabinet Knobs etc. All EXTRA. £38.00 All parts including P.C.B and sundries. £ 4.99 Set of parts including 3 P.C. Boards £22. 00 Additional components if Lummant used with Consonant (Packs.*.) £ 2.28 ■£ 1N4148 DIODES BY ITT TEXAS 100 tor £1 50 Please nole these are full spec devices Texas TIS 88A V H F F E T 10 lor £2 30 100 lor £20 00 555 TIMER 10 for £2 SO 741 O P Amp 1 10 for £2 00 MULLARD MOOULES LP1152 lOOp LP 1 153 400p LP1165 400p LP 1 166 400p LPH68 400p LPH69 400p LP1173 LP1181 LP1400 EP9000 EP9001 EP9002 ELC 1043/05 2Wp 2«0p 2«p DECODER BOARD CONTAINING W 18 x 74156. 2 x 74155. 2 x 7409. 1 x 74180. 1 x 74150. 1 x TIP32. 2 x 60 way Edge Connectors. Only few left of this unrepeatable bargain £3.50 eac ’VoTENTlOMCTERS IK UN 5K lOG 5K UN iCK LOG i OK 1 IN 75K LOG ?5K UN 50K lOG 50K LIN 100K tOG 100K UN 260* LOG ★ ★ LIMITED OFFER BC237 100 for CS.00 ★ ★. 500* LIN 1M LOG IMLIN TMiOG 7M UN AM at 30p Each ►— — ■ i ELEK TORNADO AMPLIFIER KIT All parts £12.00 P.C Board £2 96 Power Supply parts P O_A. L- 4 GANGED POTS AN at SOp Each 5K ♦ 5K LIN or LOG 10K • 10K L IN m LOG ?5K • 25K LIN Of LOG 50K • 50K i IN or LOG 100K • l DO* UN Of LOG 250* • 750* LIN of LOG S00* • 500* LIN of LOG 1 M • 1M UN of LOG k ?M • 2M UN o« LOG A OIL SOCKETS SPIN 1 3p 14 pin tap • 6 PIN Ife 28 pm 46* L ^ X GHt COUNTER Time base and control board parts with P.C.B. 9887 1 Counter and display boefd with P.C.B. 9887 2 L.F. Input board with P.C.B. 9887-3 £5.77 H F. Input board with P.C.B. 9887-4 £11.91 Transformer for above A complete set of parts ava labie with an P.C.B s. cabinet front panel .... . £98.00 - A Ht^i quality Trimmer Caps Mm Max 2 SpF 6pF All one 3.5pF - l3pF pres 20p 7 0pF-35pF FERRITE BEADS 6MM long OD 3MM ID 1MM 3p each 100 for £2 00 Paper Q.5uF 400V AC Caps Idea/ for Car ignition Systems ate 50p tedh Aaortad Japan* ae I.F. Tramformers 20 for £1.26 FERRITE RINGS FX1593 0 0. 12 m.m. I.D.6m.m 10 for 7 Op • • • • • XT Al MIC inserts 75p each 5' Scopetubes SE5J (for callers only) £15 00 eech Bias Rejector ceils 50 iOOkhZ 66,1 “4 • • • • 1 MHz CRYSTALS £3.00 Push to Make Switches 2 Op each Chokes lOuH 35p *ach lOOuH 65p each Futaba 5LT02 Non Multiplexed 4 Digit Phosphor D.ode Display With AM'PM Colon £5.00 PRE SET POTS MULLARD POT CORES LA3 IOOSOOXmZ 76* LA4 10 30KMZ 100* L AS 30 IOOKHZ LAI 3 TOOp B2Y88 Zener Diodes lOp each or 100 assorted tor £6.50 1 AMP 50V 1 AMP tOOV 1 AMP TOOV 1 AMP 400V l AMP 900V 1 AMP 1000V 7 AMP 50V 2 AMP 100V 2 AMP TOOV CASSETTE INTERFACE 9905 All parts including P.C. Board £15.00 MICRO BLOCK 2102 250 Nano Sec Static RAM H024 x 1 BIT) £2 20 each 4 for £8.40 6 for £16.00 fa 2102 450 Nano Sec Stale RAM (1024 x 1 BlTl £1.00 each 2112 450 Nano Sec Static RAM (256 x 4 BIT) 251 3 Character Generator Upper Case £7.00 eech fa 251 3 Character Generator tower case £7 0© eech fa MM5204 6 Rom £1.00 eech fa 8212 8 Bit m/out . Port 0.00 eech fa 8080 an MRU (CPU) £12.00 each. 8831 Tn State Line” Driver £2 00 eech 8833 Tri State Tram Transceiver £2 00 eech 8835 Tri State £2.00 ae«* AY5-1013 U/ART £8.00 T. POWELL 306 ST PAULS ROAD HIGHBURY CORNER LONDON N1 01-226 1489 BARCLAYCARD SHOP HOURS MOW - FRIO - 5 30 P SAT 0 - 4.30 P Christmas holidays closed: - Dec. 21st — Jan 2nd. PRICES valid at the time of going to press ALL PRICES INCLUDE POST AND VAT