ELECTRONICS WORLD · 2019-07-17 · 111111111111111[11 11111 Ill 11111111111111111111 t rrupt m/NIor e, pan a Outpui-- Included a Standard-3 & t4 din Dot matrix. 1-PLaserlet and PostScript

ELECTRONICSWORLD+ WIRELESS WORLDFEBRUARY 1994 £1.95

BASSSPECIALBass technology:Bass deception?The bass race

Designer bass:Big bass -small boxCoupled cavities

RF DESIGNUsing broadbandRF amplifier blocks

BROADCASTDAB: soundtechnology for thenext century?

APPLICATIONSSingle chip GPS,inductorlessvoltage conversion

0 2>

Denmark DKr. 70.00Germany DM 15.00

Greece Dra.760Holland DFI. 14

Italy L. 7300IR £3.30

Spain Pts. 780Singapore SS 12.60

USA 56.70

SOR DISTRIBUTION

A REED BUSINESS PUBLICAI luN

9 770959 83300 4

20% OFF: Audio and HiFi HandbookSee inside for details

THE WORLDS No.1 BEST SELLINGUNIVERSAL PROGRAMMING AND TESTING SYSTEM.

The PC82 Universal Programmer and Tester is aPC -based development tool designed toprogram and test more than 1500 ICs. The latestversion of the PC82 is based on the experiencegained after a 7 year production run of over100,000 units.

The PC82 is the US version of the SunshineExpro 60, and therefore can be offered at a verycompetitive price for a product of such highquality. The PC82 has undergone extensivetesting and inspection by various major ICmanufacturers and has won their professionalapproval and support. Many do in fact use thePC82 for their own use!

The PC82 can program E/EPROM, Serial PROM,BPROM, MPU, DSP, PLD, EPLD, PEEL, GAL, FPL,MACH, MAX, and many more. It comes with a 40pin DIP socket capable of programming deviceswith 8 to 40 pins. Adding special adaptors, thePC82 can program devices up to 84 pins in DIP,PLCC, LCC, QFP, SOP and PGA packages.

The unit can also test digital ICs such as the TTL74/54 series, CMOS 40/45 series, DRAM (evenSIMM/SIP modules) and SRAM. The PC82 caneven check and identify unmarked devices.

Customers can write their own test vectors toprogram non standard devices. Furthermore itcan perform functional vector testing of PLDsusing the JEDEC standard test vectors createdby PLD compilers such as PALASM, OPALjr,ABLE, CUPL etc. or by the user.

The PC82's hardware circuits are composed of 40set pin -driver circuits each with TTL I/O control,D/A voltage output control, ground control,noise filter circuit control, and OSC crystalfrequency control. The PC82 shares all the PC'sresources such as CPU, memory, I/O hard disk,keyboard, display and power supply.

A dedicated plug in card with rugged connectingcable ensures fast transfer of data to theprogrammer without tying up a standard parallelor serial port. Will work in all PC compatiblesfrom PC XT to 486.

The pull -down menus of the software makes thePC82 one of the easiest and most user-friendlyprogrammers available. A full library of fileconversion utilities is supplied as standard.

The frequent software updates provided bySunshine enables the customer to immediatelyprogram newly released ICs. It even supportsEPROMs to 16Mbit.

Over 20 engineers are employed by Sunshine todevelop new software and hardware for thePC82. Not many competitors can boast of similarsupport!

Citadel, a 32 year old company are the UK agentsand service centre for the Sunshine range ofprogrammers, testers and in circuit emulatorsand have a team of engineers trained to givelocal support in Europe.

* More sold worldwide than any other ofits type.

* UK users include BT, IBM, MOD, THORNEMI, MOTOROLA, SANYO, RACAL

* High quality Textool or Yamaichi zeroinsertion force sockets.

* Rugged screened cabling.* High speed PC interface card designed

for use with all PC models from XT to486.

* Over 1500 different devices (includingmore than 100 MPU's) supported.

* Tests and or identifies a wide range oflogic devices.

NOW SUPPLIED WITH SPECIAL VALUEADDED SOFTWARE (worth over £300 ifbought seperately):

* MICROTEC disassemblers for Z8, 8085,8048. 8051, 6809 & 68HC11.

Our stocked range of own manufactured andimported Sunshine products include:

* Super fast EPROM Erasers.* 1, 4 & 8 gang EPROM 8Mbit production

programmers.* Battery operated portable EPROM

programmers.* "In circuit" Emulators.* Handy pocket IC testers.

ORDERING INFORMATION

PC82 complete with interface card,cable, software and manual only I.OUD

Please add £7 carriage (by overnight courier) forUK orders, £20 for export orders, and VAT whereapplicable.

ACCESS, MASTERCARD, VISA or CWO.Official orders are welcome from GovernmVrtbodies & local authorities.

Free demo disk with device list available.

* Software supplied to write own testvectors for custom ICs and ASICs etc.

* Protection circuitry to protect againstwrong insertion of devices.

* Ground control circuitry using relayswitching.

* One model covers the widest range ofdevices, at the lowest cost.

* No need to tie up a slow parallel port.

* Two year free software update.

* Speed optimised range ofprogramming algorithms.

* NATIONAL SEMICONDUCTOR OPALjrPAL/PLD development software.

* BATCH SOFTWARE for productionprogramming.

CPCITADEL PRODUCTS LTDDEPT. WW, 50 HIGH ST.,EDGWARE, MIDDX. HA8 7EP.

Phone now on: 081 951 1848/9

Acmes

VISA

CONTENTS

FEATURES

BASS FROM ALL ANGLESSounds in the lower octave are as rare intraditional Western music as they are innature. Of late however, pop concerts hc,veregistered on seismometers. Starting cmp100, four features show not only how t3build for better bass but also allow forpsychoacoustical effects.

EXTENDING BASS 100Coupled -cavity enclosures are ideal for subwoofers butthey need careful design. Ian Gosling discusses coupled -cavity theory and presents a worked design example.

BIG BASS FROM A SMALL BOX 107Small loudspeaker enclosures can be forced to reprodicethe deep base. Jeff Macaulay's reflex system delivers :hebottom registers through acoustic and electroniccompensation.

THE SPIRIT OF BASS 112Ben Duncan looks at the psycoacoustics of bass ancdiscusses the multitude of reasons why U2 in your iv ngroom can only be a sad reflection of the real thing.

THE BASS RACE 117Speaker makers continually strive for better If respcnsefrom ever smaller enclosures. Acoustic design consul=tDavid Berriman looks at the features and compromises ofloudspeaker enclosures and driver electroacoustic interfaces.

SLOPING VEES FOR VHF 123For VHF and lower UHF work, the sloping vee antenna ishardly ever on the options list - unjustly according toRichard Formato. His claim is supported with both designinformation and performance figures.

DISTORTION IN POWER AMPLIFIERS 137Douglas Self explains the critical design features of thefeedback network. He also presents a fully -worked designdelivering 0.0015% THD at 50W.

DAB - DELIVERY, DELAY OR DEBACLE? 160Technologically, Digital Audio Broadcasting can offermore radio channels and significantly better reception.Norman McLeod discusses both the technology and itsimplications for local and national broadcasting.

USING RIF TRANSISTORS 170Designing a linear rf amplifier to perform adequately overa wide temperature range demands accurate bias. NormDye and Helge Granberg explain the design points oftemperature compensated bias circuits .

FRONT COVER BOOK OFFERYour opportunity/ to buy the neW Audio and Hi-Fi Handbook from Newnes at 20% discount.Use the special reply card located betweenpages 144/5.

REGULARS

COMMENT .....91Unsound ideas at the European Commission.

UPDATE 92Ultra -fast SiGe transistors for new radio front ends, CDlife threat, Bulletin boards and copyright, Majorcompanies in MPEG 2 scramble, 64Mbit ram jointventure, War on electronic counterfeiting.

RESEARCH NOTES 97New technique promises high -density storage, better p -types for cmos on collision course with Jupiter, proof ofpower -line cancer?

LETTERS 125Student plea, distortion in power amplifiers, coaxialdielectrics debate, scientific theory, radio hams react,distortion on AM, what happened to allophone speechchips?

PC ENGINEERING 127Allen Brown traces the progress of Mathcad to its latestupgrade - version 4 - which has been rewritten for 32bit operation. Was it worth it, he asks?

DESIGN BRIEF 130As yet, rf building blocks are not as easy to implementas their digital -circuit counterparts. But broadbandamplifier chip promise to make the job easier.

CIRCUIT IDEAS 146Smart fet battery charger, twin frequency markergenerator, A -to -D cards for PCs, radiation detector,Wien oscillator has single -pot tuning, inductivelyisolated data link.

NEW PRODUCTSThis month "s selection of recent product launchesranges from ceramic resonators to design software.

APPLICATIONS SUMMARY.. 156Global positioning core in one [C, 2.5A switching froma SOT23 transistor, modified doubler for regulated,intermediate voltages, fast, high-performance sampling,applying the fastest op -amp.

In next month's issue: Working with realtime computing. There is- or should be -a world of difference between the machines andsoftware which handle business applications and the sort whichcontrol a production line or a scientific experiment. OS -9 starts here.Also in this issue: Using transconductance amps.

MARCH ISSUE ON SALE FEBRUARY 24.

February 1994 ELECTRONICS WORLD + WIRELESS WORLD 89

Welcome tothe world of DSP.

"A Simple Approach to Digital Signal Processing" is Texas Instruments'new public Mon for anyone who needs to know about DSP but whodoesn't hive the first idea what it's all about.

Ideal 1.or students E- nd mature engineers who want to get into thed of DSP, t lis book gives a very readable introduction to

highly terhmcal subjects such as sampling, filters,freq-_ ency transforms, data compression and design

thcisions.Unlike other DSP publications, the book is

light on complicated mathematics and heavy ondiagrams, examples and clear explanations. It

also includes a typical development cyclefor engineers who need to design and

build a DSP system.Order your copy now and open up a

new world of DSP designs. Just mail thecoupon below with your cheque to:

Texas Instruments, Black Horse Road,London SE8 5NH,Fax 81 694 0099, Tel. 81 691 9000

DIGITAL

RPMMA MAWS A 71AI.A.11114

!,4.;,.,,

Name

CompanyAddress

Postal CodeCountry

TEXASINSTRUMENTS

This is my order for A Simple Approach to DigitalSignal Processing. My cheque for is enclosed.

Title

City

Telephone

CIRCLE NO. 102 ON REPLY CARE.

eet,

ulr:enneorivty '.rtY

tic rtrinoltte rated on -the -fly tioraq as ca

editingen ive tcnponentfiased pow

111111111111111[11 11111 Ill 11111111111111111111

t rrupt m/NIor e, pan a

Outpui- - Included a Standard- 3 & t4 din Dot matrix. 1-PLaserlet and PostScript

Pent -Ai -Mere HF, GraphIsc, Roland & HoustonPhobpkitters - All Gerber 3X00 arc 4X00ExoMl.ri NC Onl / Annotated drill orawirtgs (BM2)

Call for info or fullevaluation k

Tsien (UK) LimitedTel (0354) 695359 -44110P I

Fax (0354) 69E957 tsientam (UK) Ltd, Arattby Mouse. Wormy Road, Ch$ttaral. Carabildge PEI& 6UT

90

CIRCLE NO. 103 ON REPLY CARD

ELECTRONICS WORLD+WIRELESS WORLD February 1994

COMMENT

EDITORFrank Ogden

081-652 3128

DEPUTY EDITORMartin Eccles081-652 8638

CONSULTANTDerek Rowe

DESIGN & PRODUCTIONAlan Kerr

EDITORIAL ADMINISTRATIONLorraine Spindler

081-652 3614

ADVERTISEMENT MANAGERCarol Nobbs

081-652 8327

SALES EXECUTIVEPat Bunce

081-652 8339

ADVERTISING PRODUCTIONPaul Burgess

081-652 8355

PUBLISHERSusan Downey

EDITORIAL FACSIMILE081-652 8956

CLASSIFIED FACSIMILE081-652 8931

SUBSCRIPTION HOTLINE0622721666Quote ref INJ

SUBSCRIPTION QUERIES0444 445566

NEWSTRADE DISTRIBUTIONENQUIRIESMartin Parr

081 652 8171

BACK ISSUESAvailable at f2.50

081-652 3614

ISSN 0959-8332

MI REEDWit BUSINESS

PUI3LISHING

Unsound ideas at theEuropean CommissionK ()boil, doubts for a moment the

N ingenuity and technical excellenceof the BBC's digital audio broadcastsystem - DAB. Recent demonstrationsboth here and overseas show it to offerreal improvement on standard FM formobile reception. Although it could beargued that the London venuedemonstrations were flawed by therelative siting of the FM and DABtransmitters sending out the testmaterial - the DAB sites were localwhile the FM site was 20 miles away -it eliminates multipath distortion andco -channel interference. DAB providesan excellent transmission medium formobile reception. Why then should wethink very carefully before committingourselves to a new sound broadcastreceiving system?

It is simply this. The technology,while feasible, takes little account ofbroadcasting requirements. It has beentailored far too closely to therequirements of a national network withlittle account of local broadcasting.

The precise details of the technologyare dealt with elsewhere in this issue butin essence, each DAB transmitterbroadcasts six programmessimultaneously using subcarrierinterleave. The frequency spreadingreduces the individual data rate/Hz tothe point where multipath is no longer aproblem but it requires that sixprogrammes are transmittedsimultaneously from a single site.Where used for local broadcasting, itwould mean that six stations are lockedtogether into an inflexible bundle. Sixfranchises would have to be offered to

serve a local area since DAB onlyrepresents efficient use of frequencyand financial resource when fullyoccupied. While this arrangementclearly suits Radios I to 5 plus another,it leaves local radio out in the cold.

The European Commission feelscompelled to push Eureka DAB to gainadvantage over emerging UStechnology in setting world standards.American digital sound broadcasttechnology takes as its starting point theelimination of transmissionshortcomings from individual radiostations. And most broadcast systemsaround the world operate like theAmericans'. I don't argue withassistance to our home grown DAB onphilosophical grounds: let's gain anytrading advantage we can. However, theworld will surely turn its back onanything which isn't absolutely in tunewith a market requirement.

Perhaps I should remind the EU of theMAC TV standards fiasco. And alsothose at the BBC/ITC with similarlyshort memories.

No audience or broadcastingorganisation can afford to supportmultiple standards on the airwaves,particularly when the actual cost ofhardware for the alternative will bequite high in comparison to theequipment which it is designed toreplace. I urge the EU and our broadcastR&D departments to look again at thewhole problem, not just selective partsof it. A total solution will receive awider audience than a simply pragmaticone.

Frank Ogden

Electronics World + Wireless World is published monthly.By post. current issue £2.25. back issues (if avaiable)£2.50. Orders. payments and general correspondence toL333, Electronics World + Wireless World, QuadrantHouse, The Quadrant, Sutton, Surrey SM2 5AS.Telex:892984 REED BP G. Cheques should be madepayable to Reed business Publishing Group.Newstrade: IPC Marketforce, 071 261-5108Subscriptions: Quadrant Subscription Services. OaklieldHouse. Perrymount Road, Haywards Heath. Sussex RH163DH. Telephone 0444 445566. Please notify a change ofaddress. Subscription rates 1 year (normal rate) £30 UK and£35 outside UK surface mail. Air mail rates are found on cardbound into issue.USA: $78.00 airmail. Reed Business Publishing fUSA).

Subscriptions office, 205 E. 42nd Street, NY 10117.Overseas advertising agents: France and Belgium: PierreMussard. 18-20 Place de la Madeleine,Paris 75008. UnitedStates of America: Ray Barnes, Reed Business PublishingLtd. 205 E. 42nd Street, NY 10117.Telephore (212) 867-2080. Telex 23827.USA mailing agents: Mercury Airfreight International LtdInc, 10(b) Englehard Ave, Avenel NJ 07001. 2nd classpostage pad at Rahway NJ Postmaster. Send addresschanges to above.Printed by BPCC Magazines (Carlisle) Ltd, NewtownTrading Esiate, Carlisle, Cumbria. CA2 7NRTypeset by Marlin Graphics 2-4 Powerscroft Road, Sidcup.Kent DA14 5DT©Reed Business Publishing Ltd 1992 ISSN 0959 8332


UPDATE

Ultra fast SiGe transistorsgo commercialThe world's first highly integrated silicon-

! germanium semiconductors will be onthe market within nine months following anagreement between Analog Devices andIBM. Analog Devices plans to design arange of ICs for radio frequency and mixedsignal applications, which will be made byIBM in New York.

The new production process, which usesultra high vacuum chemical vapourdeposition to create a Ge on Si structure,allows the building of transistors with an fTof more than 60GHz without extraprocessing steps. SiGe devices can operatefrom 3 and 1.5V supplies.

Silicon -germanium semiconductors areseen as the natural successor to cmos forhigh speed circuits. A I GHz 12 -bit digital toanalogue converter will be the first IC.

NTSC to VGA conversion: ITT's digital TVcomponents perform a TV to computerdisplay conversion for multimedia. The videoadaptor uses an NTSC comb filter chip, theDPU2554 deflection processor, progressivescan processor PSP2210 and video codecVCU2134.Simple multimedia applications do notrequire data reduction. The picture data isfirst converted to the square pixel format ofcomputers. These are then interpolated to therequired number of pixels for display on thecomputer screen using a picture formatprocessor chip DTI2250.

However, Analog Devices and IBM arelikely to concentrate on developing single ICfront end chips for wirelesscommunications.

This is a significant blow to the fortunes ofgallium -arsenide ICs - previously the onlyalternative semiconductor process for veryhigh speed circuits. SiGe circuits aresignificantly cheaper to produce and can beintegrated onto an ordinary cmos process.

Richard Siber, director for the wirelesscommunications service at marketresearcher BIS Strategic Decisions, said:"This process will negate the need for moreexpensive GaAs for operating frequenciesup to 3GHz. This development is trulyrevolutionary, giving the wireless industry amajor boost."

See also Research Notes, p97

Jessi goescommercialjessi, the European microelectronicsresearch and development programme, is

to produce a series of GSM chips for pockettelephones working on the European digitalcellular network. These are an ATM chipsetfor advanced data communicationsequipment, a chipset for receiving digitalaudio broadcasts (DAB), a chipset providingthe electronics for digital TVs, and anautomotive safety chipset.

If used commercially, the chips could giveEuropean systems manufacturers an edge inthe market for end equipment. Theyrepresent a shift away from the previousobjective of Jessi which was to push forwardtechnological capability over variousdisciplines without the specific intention ofproducing useful, commercially competitiveproducts.

The ATM project is the most advanced.Project leader, Alcatel, already has the four -chip set in silicon on a 0.8µm process andaims to put them onto the Jessi developed0.5µm process by 1995.

The other three projects are less advanced.A two -chip GSM chipset has been designedby Alcatel, Bosch, SGS-Thomson andMietec but is some way fromimplementation in silicon.The chipsetfollows the industry pattern of one chip forthe radio frequency side and another for thedigital signal processing side. The digitalTV, DAB and car safety chipsets are stillbeing designed

Bulletin boardsface copyrightbattle

Amusic publisher in the US is suingCompuServe for copyright

infringement on one of its bulletinboards. The case, to be heard in theFederal Court of New York, looks likelyto set far-reaching precedents. By suingCompuServe, as provider of the hostcomputer on which the music material istemporarily stored rather than individualsystem users, the music publishers arecreating a precedent for the future. Allelectronic delivery services couldbecome legally responsible for whatevermessages subscribers post through them.

The Harry Fox Agency, part of theNational Music Publishers Association,is paying for music publisher FrankMusic to pursue the test case on behalfof over 140 other publishers.CompuServe describes itself as "theworld's most comprehensivecomputerised information service". Theclaim is for $70 million in damages andcosts for copyright infringement in justone song. If the publisher wins, theHFA, which represents 12,000publishers and controls the licensing of75% of all the music played in the US,will claim from any other electronicdistribution system which carries music.

Over a million PCs around the worldconnect to the CompuServe network toexchange electronic mail messages andaccess 1700 different bulletin boards.Most of the messages are text, forinstance news and views on newtechnology. But one board, called theMIDI/Music Forum, lets subscribersexchange music.

Because midi code is similar to ascii,subscribers to CompuServe have beenconverting music into digits anduploading it into the CompuServecomputer so that other subscribers candownload the code to make a PC orelectronic instrument play the tune.

By logging use of the musical bulletinboard, the HFA has been able to citewhat it describes as 690 "wilful acts ofinfringement", involving more than 500songs owned by some of the 12,000music publishers it represents. BF

92 ELECTRONICS WORLD + WIRELESS WORLD February 1994

UPDATE

Co-op ventureproduces 64Mbit ramTexas Instruments and Hitachi haveI unveiled the 64Mbit dram they have

been developing for the past two years.The 228mm2 chip can store four copiesof Gone with the wind - some 2000printed pages.

Dr Tsugio Makimoto from Hitachi said:"We pooled the best of the technologiesof each side to realise benefits in terms oftechnology and development efficiencythat greatly exceeded what either wouldhave been able to achieveindependently."

He reckons the success of the project"provides a platform for substantiallystrengthening and expanding our co-operative efforts in this area".

The firms are considering a similarproject for the 256Mbit dram. Toshiba,IBM and Siemens have already startedjoint 256Mbit development.

A change from TI's previous drams isthat the chip uses stacked rather thantrench capacitor techniques.

Beeb in more HDTV tests

TheBBC and Thomson-CSF

Laboratories recently carried out asecond series of HDTV transmissions fromCrystal Palace to test reception quality in awide area around London. Transmitterpower was 10kW and channel 34 was usedwith an 8MHz bandwidth, the same as anordinary Pal signal.

Bit rate was 30Mbit/s and thecompression system used a motioncompensated hybrid discrete cosinetransform similar but not identical to theMPEG-2 format.

There were 500 carriers using anorthogonal frequency division multiplexand 64 -state QAM modulation.

Philips claims to the first company to produce a60W, single ended output device for use in theband 1.5 to 1.7GHz. The LFE15600X npntransistor achieves a typical power gain of 8.5dBat 1.5GHz and has been designed for class ABlinear service: Intermodulation products are30dB down at 60W pep output.

Cheap packaging threatens CDsThe latest scare story about di i-itegrating

CDs appearing in the UK press may havebeen hyperbolic in its predictions of millionsof CDs doomed to disintegrate but the factsbehind the story have substance. The causeis not in the CD standard or manufacturingprocess, but sulphur content of thepackaging.

Although the CD standards set by Philipsspecified sleeve size they did not specifymaterials. Analysis of the sleeves suppliedby the record companies showed that somewere made from high quality processedpaper called solid sulphite board, and did notaffect any discs, other sleeves were madefrom untreated wood pulp. This releasessulphur compounds which eat through theprotective lacquer on the label surface of theCD to erode the metal reflective layerunderneath.

The first proof that CDs would not lastforever came in 1988. Pressing plantsdiscovered that the inks they were using toprint label information direct onto the discwere eating through the protective lacquercoating and destroying the very thin layer ofaluminium which reflects the laser light. Sothe discs stopped playing. Before that, plantspressing 30cm laser video discs had foundthat the glues used to stick the two halves ofthe double -sided discs together were eatingaway the reflective layer. The factoriessolved these problems by changing theirinks and glues.

Now Philips' plant in Blackburn has foundthat some CD singles pressed in the lateeighties for record companies Polygram andA and M are failing. Investigations haveidentified the cause as release of sulphurfrom some of the cardboard sleeves used topack CD singles. But the concernedorganisations do not agree on the basicissues which might place CDs made byother plants at risk.

Philips built the Blackburn PdO plant topress 30cm video discs, but demand wassmall so it began pressing 12cm CDs. It usesa wet process, similar to that used formaking mirrors, to deposit a layer of silveras the reflective surface. Most CD pressingplants sputter aluminium but the necessaryequipment was not available when theBlackburn factory was built.

In 1988 the record companies started tosell CD singles, full size I2cm discscarrying only around 20 minutes of music.They cut costs by using cardboard sleevesinstead of plastic jewel boxes. In late 1989the plant found that some discs in boardsleeves were refusing to play after a few

months use, while others played perfectly.To confuse the issue, some discs would playon some players but not others, and thenlater fail on all players.

Cheap CD players immediately refused toplay the disc while more expensive playerswith better error correction circuitry playedthe disc perfectly, but failed after severalmore months when the holes had growntarget.

The two types of sleeve look the same."We had a real struggle to distinguishbetween them, but finally developed asimple test" says Dave Wilson, PdO'sTechnical Services Director. PdO found thatif a drop of methylated spirits is put on theboard surface, the material turns clear toreveal a pulp of free fibres if the board isuntreated. PdO then worked with Philips toset a standard for CD sleeve chemistry.Although some record companies hadstarted to use plastic jewel boxes for singles,because of the perceived low value of card,the card sleeve is still used in somecountries, including France. Recordcompanies in the US use card sleeves forfull length CDs.

Polygram believes that the problem wasconfined to the Blackburn factory becausethe silver is more susceptible to sulphur thanaluminium; the effect is similar to silvercutlery tarnishing. The disc gradually turnsfrom silver to bronze and loses itsreflectivity. But Dave Wilson of PdObelieves that aluminium will degrade inexactly the same way if the recordcompanies supply any plant with card whichhas a high sulphur content.

PdO now checks all card for sulphurcontent. PdO also checks the paper inlaynotes which sit inside a jewel box and pressagainst the disc lacquer. Dave Wilson saysthat all PdO technical information on thesulphur risk was made available to otherplants. But the record divisions of EMI,A & M and Polygram appeared unaware ofthe need to use high quality paper and boardfor the CD sleeves and inlays which theysupply to pressing plants. EMI says it isconfident that the problem is confined todiscs pressed by PdO. But a spokeswomanfor EMI's own CD pressing plant inSwindon was unaware of the need to checkpaper and card for sulphur content.

It now seem only a question of time beforesomeone, somewhere, reports full lengthCDs rotting because they have either beenpackaged in contaminated card boxes orpacked with contaminated paper sleevenotes. Barry Fox


UPDATE

Major companies in MPEG 2 scramblenticipating profit in digital TV, AT&T

rlMicroelectronics, LSI Logic and SGS-Thomson Microelectronics have allannounced details of MPEG-2 compliantreal-time video decoders. Theannouncements presage the arrival in 1994of digital TV receivers for satellitebroadcasts and cable TV set -top decoders.

AT&T demonstrated its chip, the AV6I 01,late last year at the Western Cable Show inthe US. The decoder chip was paired with areal-time encoder system developed byAT&T Transmission Systems.

Anne Schowe, managing director of

AT&T's visual solutions business unit, said:"We are claiming victory in the scramble todeliver MPEG-2 video decoder chips to thedigital video market. Our AV6I01 chipdecodes all -the MPEG-2 video layers in realtime without requiring external processorsupport, which means it's an ideal solutionfor inexpensive set -top decoder systems."

The IC is expected to cost less than $100in volume.

But AT&T's claim is disputed by SGS-Thomson: "We would question the AT&Tclaim that its chip is the first real-timeMPEG-2 decoder chip to be demonstrated.

War opens on electronic counterfeitM ational Westminster Bank's says it has

N "high confidence" in the security of itsnew electronic alternative to cash, theMondex card. NatWest has been working onthe smart card cash system for nearly fouryears, hopes it will become a globalstandard. It plans to introduce the system in1995.

In its basic form Mondex relies on aconventional smart card, made to ISOstandard 7816, with inbuilt memory chipand computer processor which store cashcredits, and external contacts to connectwith a card reader. Anyone can use the cardto make a purchase from any shop which hasa reader at the till. The user's only securityis to use an electronic wallet to lock, andunlock, the card's memory with a personalidentification number.

The conceptual breakthrough claimed byNatWest is in the method of proofing thecard against counterfeiting, so that criminalscannot make copies of cards or tamper withthe memory and credit transfer signals. TimJones, NatWest's Head of InformationTechnology Policy and Strategy, saysMondex is "extraordinarily secure". Butcourt actions brought recently by satellitebroadcaster BSkyB reveal that smart cardscan be a lot less secure than those who relyon them previously thought.

Credits are loaded into the Mondex card

memory or deducted to make payments by areader at the point of sale. Both NationalWestminster and Midland Bank will providehole -in -the -wall readers which letcardholders refresh cards by loading creditsinto the memory, while debiting theiraccounts. BT is already designing adomestic "smart" telephone which will letcard -holders refresh them from home bycalling their bank.

NatWest knows that Mondex is an openinvitation to criminals who will try to printmoney by copying cards or pirating thesignals which transfer cash from onememory to another. A spokesman for thebank "Yes we are definitely confident onsecurity. We realise that Mondex will betargeted by criminals. There are many levelsof protection against counterfeiting".

The system checks the integrity of themoney signal passing from "purse" to"purse", or source and destination, to ensurethat a card -holder does not tamper with thedigits and so make a transfer of 10 registeras £1000. Mondex also checks that eachsignal only registers once, to stop the same£10 transfer notching up five times tobecome £50. The system continually checksthe validity of each purse, to ensure that theowner of one card cannot suck money fromsomeone else's account. BF

Picture of the US electronics industry: normalised, composite graph from Electronic BusinessForecast leading indicator demonstrating year on year change in the production, order books, shareprice, semiconductor bookings and interest rates. Contrary to common perception, the cyclicbusiness swings appear to be smoothing out as industry learns to plan for the perturbations.

160

140

120

100

80

vclicalhigh

cyclicalow

cyclicallow

y, licalhh,.h

cych'allow

cyclicalhigh

source Electronic 1111tinen for...st

608182 83 84 85 86 87 88 89 90 91 92 93 94 95

We have been privately showing ours, theSTi3500, for a while," said Simon Loe thefirm's technical spokesman.

The AT&T chip provides a 4:2:2 rasteroutput and needs 8Mbit of external memoryas a frame store. It uses a 27MHz clock andconsumes 1.3W.

The chip's i/o operates from a 5V supplywhereas the core uses a 3.3V supply. Initialsamples are being provided to a fewcustomers but production is not scheduleduntil later this year.

AT&T says the AV610I device is a firstgeneration chip capable of decodingCCIR601 broadcast quality resolutionpictures conforming to the main level simpleprofile format in the MPEG-2 specification.This format excludes bidirectionallypredicted, or B frames, in the picturesequence. They are computation intensive toencode and decode.

Main level, main profile format picturesneed a minimum of 16Mbit for framestorage.

A second generation chip is planned thatcan decode a main level, main profile datastream including B frames. However, AT&Tbelieves in the short term system builderswill opt for the cheaper main level, simpleprofile format.

In contrast the STi3500 chip canaccommodate main level, main profile andmain level, simple profile picture formats byvirtue of a programmable frame store. Thechip can directly support between 8 and32Mbit of dram. An external microcontrolleris needed.

Martin Bolton, technical marketingmanager for the image compression group,says the external microcontroller is not adisadvantage: "There will usually be amicrocontroller somewhere in the systemand the load required to control the STi3500is small. It is only a matter of settingregisters every picture."

Also allowing the decoder to be controlledat the frame and field level externally allowsa greater calltibility for handling multiplestandards and, importantly, special modes,said Bolton.

The STi3500 incorporates all the decoderfunctions, inverse discrete cosine transform,inverse quantise, frame predictionconstructor and variable length decoder. Thechip is highly pipelined to attain thenecessary performance. The output digitalvideo signal is in a YCrCb formatmultiplexed onto an 8bit bus. Thesynchronisation signal has to be suppliedexternally.

The LSI Logic L64000 video decoder willalso work with a main level, main profilepicture format and has been developed withZenith Electronics - one of the companiesinvolved in the US grand alliance to developan HDTV standard.

Simon Parry, Electronics Weekly.


AMSTRAD DIVDP4000 Entire printer assemblies includingprinthead, platen, cables, stepper motors etc. Everything bar the electron-ics and case. Good stripper!! Clearance price just £5 REF: MACS or 2for £8 REF: MACSVIEWDATA SYSTEMS Brandnew units made by TANDATAcomplete wlh 1200/75 built in modern, infra red remote controledkeyboard. BT approved, Prestel compatible, Centronice printer pertRGB odour and composite output (works with any TV) complete withpower supply and fully cased. Price is just £20 REF: MA020 Alsosome customer returned units available at E10 each REF: MAG10PPC MODEM CARDS. These are high spec plug in cards madefor the Amstrad laptop computers. 2400 baud dial up unit completewith leads. Clearance price is £5 REF: MAG5PIINFRA RED REMOTE CONTROLLERSOtiginalymadeforhi spec satellite equipment but perfect for all sorts of remote controlprojects. Our clearance pnce is just £2 REF: MAG2TOWERS INTERNATIONAL TRANSISTOR GUIDE. Avery useful book for finding equivalent transistors, leadouts, specsetc. E20 REF: MAG20P1

SINCLAIR C6 MOTORS We have a few left without gearboxes.These are 12v DC 3,300 rpm 6'x4'. 1/4'OP shaft. E25 REF: MAG25

UNIVERSAL SPEED CONTROLLER KIT Designed by usfor the above motor but suitable for any 12v motor up to 30A.Complete with PCB etc. A heat sink may be required. £17.00REF: MAG17VIDEO SENDER UNIT. Transmits both audio and video signalsfrom ether a video camera, video recorder, TV or Computer etc toany standard TV set in a lay range! (tune TV to a spare channel) 12vDCop Priceis£15 REF: MAG15 12v psu is£5extra REF: MAG5P2FM CORDLESS MICROPHONE Small hand held unit with a500' range! 2 transmit power levels. Reqs PP3 9v battery. Tuneableto any FM receiver. Price is £15 REF: MAG15P1LOW COST WALKIETALKIES Pair of battery operated unitswith a range of about 200'. Ideal for garden use or as an educationaltoy. Price is E8 a pair REF: MAG BPI 2 x PP3 req'd.

MINATURE RADIO TRANSCEIVERS A pair of walNetalkies with a range of up to 2 itildnetres in open country. Unitsmeasure 22152x155mm. Complete with cases and earpieces. 2x PP3req'd. £30.00 pair REF: MAG30COMPOSITE VIDEO KIT. Converts composite video intoseparate H sync, V sync, and video. 12v DC operation. £8.00REF: MAG8P2.LQ3600 PRINTER ASSEMBLIES Made by Amstrad they areentre mechanical printer assembles including printhead, steppermotors etc etc In fact everything bar the case and electronics, a goodstripper! £5 REF: MAG5P3 or 2 for £8 REF: MAG8P3PHILIPS LASER 2MW helium neon tube. Brand new full spec£40 REF. MAG40. Mains power supply kit £20 REF: MAG20P2.Fully built and tested unit £75 REF: MAG 75.

SPEAKER WIRE Brown two core, 100 foot hank £2REF: MAG2P1

LED PACK of 100 standard red 5mm leds £5 REF: MAG5P4JUG KETTLE ELEMENTS good general purpose heatingelement (about 2inv)ideai for allsorts of heating projects etc 2 for£3REF: MAG3

UNIVERSAL PC POWER SUPPLY complete with flyleads,switch, fan etc. Two types available 150w at £15 REF:MAG15P2(23x23x23mm) and 200w at E20 REF: MAG20P3 (23x23x23mm)OZONE FRIENDLY LATEX 250m1 bottle of liquid rubber, setsin 2 hours. Ideal for mounting PCB's, fixing wires etc E2 each REF:MAG2P2 FM TRANSMITTER housed ina standard worldng 13Aadapter!!the bug runs directly off the mains so lasts forever! why pay £700? orpnce is £26 REF: MAG26 Transmits to any FM ratio.' FM BUG KIT New design with PCB embedded cot for extrastability. Transmits to any FM radio. 9v battery read. £5 REF.MAG5P5 FM BUG BUILT AND TESTED superior design to kit, assupplied to detective agencies etc. 9v battery req'd. £14 REF:MAG14TALKING COIN BOX STRIPPER originaly made toretail at E79 each, these units are designed to convert and ordinary

phone into a payp hone. The units we have generally have the lodesmissing and sometimes broken hinges. Howeve rthey can be adaptedfor their original pupose or used for sometNng else?? Price is just £3REF: MAG3P1

100 WATT MOSFET PAIR Same spec as 2SK343 and2q4413(8A,140v-100w) 1Nchearteler10-1.Pcbm1, E3 a pair REF:MAG3P2VELCRO 1 metre length of each side 213mm widefixing for temporary jobs etc) £2

'4AVAKEDCAGIT Cosisting of a cased mains motorwith lead. The motor has two magnets fixed to a rotor that spin roundinside. There are also 2 plastic covered magnets supplied. Made forremotely sir ning liquids! you may have a use?£3 each REF: MAG3P32 for £5 REF: MAG5P6TOP QUALITY SPEAKERS Made for HI Fl televisions theseare 10 watt4RJap made 4' round with large shielded magnets. Goodquality general purpose speaker. E2 each REF: MAG2P4 or 4 for E6REF: MAG8P2TWEETERS 2' diameter good quality tweeter 140R (would begood with the above speaker) 2 for E2 REF: MAG2P5 or 4 for £3REF: MAG3P4AT KEYBOARDS Made by Apricot these quality keyboards needjust a small modification to run on any AT, they work perfectly but youwill have to put up with 1 or 2 foreign keycaps! Price £6 REF.MAG6P3XT KEYBOARDS Mixed types, sane retums, sane good, someforeign etc but all goal for spares! Price is E2 each REF:MAG2P9or 4 for £6 REF: MAG6P4PC CASES Again mixed types so you take a chance next one offthe pie E12 REF: MAO 12 or hvo idenlcal onesfor£20 REF: MA020P4component pack bargain 1,030 resistors +1,000 capacitors (allsame value) £250 a pada REF: MAG2P7

1994 CATALOGUEOUT NOW

MASSIVEWAREHOUSE CLEARANCE

FANTASTIC £20.00 REDUCTION

REFURBISHED PC BASE UNITSCOMPLETE WITH KEYBOARD

FROM ONLY £29.00

AMSTRAD 1512 BASE UNITSGUARANTEED

PERFECT WORKING ORDER.A LOW COST NTRODUCTION TO THE HOME COMPUTER MARKET.

AMSTRAD 1512SD1512 BASE UNIT, 5.25" FLOPPY DRIVE ANDKEYBOARD. ALL YOU NEED B A MONITOR ANDPOWER SUPPLY. WAS £49.00

NOW ONLY S29.00REF: MA G29

AMSTRAD 1512DD1512 BASE UNIT AND KEYBOARD AND TWO5.25 360K DRIVES . ALL YOU NEED IS A MONITORAND POWER SUPPLY WAS E59.00

NOW ONLY 539.00REF: MAG39

SOIAII POWER PANELS

3FT X 1FT 10WATT GLASS PANELS14.5v/ 700mA

NOW AVAILABLE BY MAIL ORDER£33.95

ewe 5100 SPECIAL PACKAGING CHARGE)

TOP QUALITY AMORPHOUS SILICON CELLS HAVE ALMOST ATIMELESS LIIESPAN WITH AN INFINITE NUMBER OF POSSIBLEAPPLICATIONS, SOME OF WHICH MAY BE CAR BATTERYCHARGING, FOR USE ON BOATS OR CARAVANS, OR ANY-WHERE A PORTABLE 12V SUPPLY IS REQUIRED. REP MAG34

ALSO 1FT X 1FT GLASS SOLAR PANELS 12v 200mAONLY E15.00. PEP: MAG15P3

FREE SOFTWARE!Brand new, UNUSED top quality Famous brandlicensed software discs. Available in 5.25' DSDD or 5.25'HD only. You buy the disk and it comes with free BRANDNEW UNUSED SOFTWARE. We are actually selling you thefloppy dkc for your own "MEGA CHEAP" storage foclatles,If you happen to get software that you want/need/like aswell you get a 'MEGA BARGAIN' tolDSDD P1C710 £2.99 REF: MAG3P7 PKT100 516.00 REF: MAG16HD PKTTO 53.99 REF: MAGIPS PKTIO0 525.00 REP: MAG26P1

LARGER GUMMY PRICES AVAILABLE ON APPLICATION

£SLUSEWE BUY SURPLUS STOCKLUILEETURN YOUR SURPLUS STOCK INTO CASH.

IMMEDIATE SETTLEMENT. WE WILL ALSO QUOTE FORCOMPLETE FACTORY CLEARANCE.

COMING SOON

1994 CATALOGUE.PLEASE SEND 42P , A4 SIZED SAE FOR YOUR FREE COPY.

MINDAUM 00.D4 ORDER L5.00 TRADE ORDERS FROM GOVERNMENT, SCHOOLS.UNIVERSITIERALPCAL AUTHORITIES WELCOME ALL GOODS SUPPLIED SUBJECT TOOUR CONDITIONS OF SALE AND maxes OTHERWISE STATED GUARANTEED FOR 30DAYS RIGHTS RESERVED TO CHANGE PRICES & SPECIFICATIONS WITHOUT PRIORNOTICE ORDERS SUBJECT TO STOCK QUOTATIONS WILLINGLY GIVEN FOR QUANTI-TIES HIGHER THAN THOSE STATED

'SOME OF OLR PRODUCTS MAY RE UNLICENSABLE IN THE UK

COMMODORE MICRODRIVE SYSTEM mini storagedevice Tor C64's 4 times faster than disc drives, 10 limes fasterthan tapes. Complete unit just £12 REF: MAG12P1

SCHOOL STRIPPERS We have quite a few of the aboveunits which are 'returns' as they are quite comprehensive unitsthey could be used for other projects etc. Let us know how many youneed at just 50p a and (minimum 10).

HEADPHONES 16P These are exVirgin Atlantic. You can have8 pairs for £2 REF: MAG2P8P ROY,. la ITV SENSORS These are smal PCB's with what looklike a source and sensor LED on one end and lots of components onthe rest dthe PCB. Complete with fly leads. Pack of 5 £3 REF: MAG:3P5 Oil 20 for £8 REF: MAGSP4FIBRIE OPTIC CABLE Made for Hewlett Packard so pretty goodstuff! you can have any length you want (min5m) first 5m E7 REF:MAG7 thereafter £1 a metre (ie 20m is £22).REF: MAGI M ax length250m.

SNOOPERS EAR? Onginall made to clip over the earpiece oftelephone to amplify the sound -it also works quite well on the cablerunning along the wall! Price is £.5 REF: MAG5P7DOS PACKS Microsoft version a 3 or higher complete with allmanuals or price just £5 REF: MAG5P8 Worth it just for the verycomprehensive manual! 5.25' only.DOS PACK Microsoft version 5 Original software but no manualshence only £3 REF: MAG3P8 5.25' only.FOREIGN DOS 3. 3-German,French,italian etc E2 a pack withmanual 5.25' only. REF:MAG2P9MOND VGA MONITOR Made by Amstrad, refurbished E49REF:MAG49C178644COLOURMONITOR MadebworkwiththeCPC464home aomputer. Standard RGB input so will woricwith othermachines.Refurbished £59.00 REF; MAG59

JUST A SMALL SELCTION of w hat we have to see more getour 1994 catalogue (42p stamp) or call in Mon -Sat 9-5.30

HAND HELD TONE DIALLERS Ideal for the control of theResponse 200 and 400 machines. E5 REF:MAG5P9PIR DETECTOR Made by famous UK alarm manufacturertheseare hi spec, long range internal units. 12v operation. Slight marks oncase and unboxed (although brand new) £8 REF: MAWSWINDUP SOLAR POWERED RADIO AM/FM radio com-plete with hand charger and solar pane! £14 REF: MAG14P1

COMMODORE 64 Customer returns but ok for spares etc £12REF: MAG12P2 Tested and working units are E69.00 REF: MAG69

COMMODORE 64 TAPE DRIVES Customer returns at E4REF: MAG4P9 Fully tested and working unitsare£12 REF: MAG12P5

COMPUTER TERMINALS complete with screen, keyboardand FS232 input/output. Ex equipment. Price is E27 REF: MAG27

MAWS CABLES These are 2 core standard black 2 metre mainscables fitted with a 13A plug on one end, cable the other. Ideal forprojects, low cost manufacturing etc. Pack of 10 forE3 REF: MAG3P8Pack of 100 E20 REF: MAG20P5SURFACE MOUNT STRIPPER Originaly made as sorneform of high frequency amplifier (men chip is a TSA5511T 1.3GHzsynthesiser) but good stripper value, an excellent way to play withsurface mount components £1.00 REF: MAGI P1.MICROWAVE TIMER Electronictimerwith relay output suitableto make enlarger timer etc £4 REF: MAG4P4

PLUG 420? showing your age? pack of 10 with leads for£2 REF:MAG2P 11

MOBILE CAR PHONE £5.99 Well almost! complete in carphone excluding the box of electronics normally hidden under seat.Can be made to illuminate with 12v also has built in light sensor sodisplay only ill umi nates w hen dark. Totally convindng! REF: MAG6P6

ALARM BEACONS Zenon strobe made to mount on an externalbell Sox but could be used for caravans etc. 12v operation. Justconnect up and it flashes regularly) £5 REF: MAG5P11FIRE ALARM CONTROL PANEL High quality metal casedalarm panel 350x165x8Omm Comes with electronics but no informa-tion. £15 REF: MAG15P4SU 5ER SIZE H EATSINK Superb quality aluminum heatsink.365 K 183 x 61mm, 15 fins enamble high heat dissipation. No holes!£9.99 REF: MAG1OP1PREMOTE CONT These are receiver boards forgarage do ng systems. You ma ve another use? £4 eaREF- A 5

L X Line output transformers believed to be-tbik, hi res colourm rs but useful for getting high voltages from low ohes! £243ecliREF: G2P12 bumper pack of 10 for E12 REF: MAG[2P3.

BULL ELECTRICAL250 PORTLAND ROAD HOVE SUSSEX

SQTmatt ORDER TERMS: CASH POOR CHEQUEWITH ORDER PLUS £3.00 POST PLUS VAT.

PLEASEAILOWT 10 -DAYS FOR DELIVERYTELICPISOMBORDREM WKLCOME -

TEL: 13273 393500loft.4

PORTABLE RADIATION DETECTOR

£49.99A Hand held personal Gamma and XRay detector. This unit contains twoGeiger Tubes, has a 4 digit LCD dis-p ay with a Piezo speaker, giving anaudio visual indication. The unit de-tects high energy electromagneticquanta with an energy from 30K eV toover 1.2M eV and a measuring rangeof 5-9999 UR/h or 10-99990 Nr/h. Sup -Oiled complete with handbook.

REF: MAG50


February 1994 ELECTRONICS WORLD+WIRELESS WORLD 95

"... there is no doubt that running underfield and makes it a visually attractive

Wireless WorldHigh Quality PCB and Schematic

0 Supports over 150 printers/plotters including 9 or

Windows puts it ahead of thepackage." Electronics World

July 1993Design for Windows 3/3.1 and DOS

''T.--,7.._ L.:,"="' -''''"J''''''' r -i-

+4:11:11)

I et

*II

=rift'et

24 dot-matrix DeskJet, LaserJet, Postscript, 1 ''' 040U:91:5i 'x'-"' I 0"." I I -1*pin ;:,60,000-50

and HPGL. Professional Edition imports GERBER Sfiles, and exports GERBER and NC DRILL files.

/7r111M.1PMMINI4

0 Up to 200,000 pads/track nodes depending on 11111111111111.0

1,4

memory. Simple auto -router and schematic capturetools SPICE

11.000., 00.G 0 IAPO 2.0 0 so

o'°°with compatible net -list output. OKiiM4 0---1---b:0-s-, 1:101:101:11:11:11E100111:1 ao s nos r 2300

0 Low cost DOS version (reduced features) also0--0 -AI

available. Ring for full details! K K rifienl IreiiiilL

"Quickroute provides a comprehensive and effective introduction to PCBdesign which is a pleasure to use" Radio Communication May 1993.

from

£39Za-Tim

POWERware, Dept EW, 14 Ley Lane, Marple Bridge, Stockport, SK6 5DD, UK.Ring us on 061 449 7101 or write, fora full information pack.

POWERwareQuickroute is available for Windows 3/3.1 in Professional (£99.00) and Standard (£59.00) editions,Software .

Design and for DOS with reduced features (£39.00). All prices inclusive. Add £5 P+P outside UK.


RF MODULES UP TO 2GHzGASFET LNAs 5MHz-2G1-1zTwo -stage. High Q filters. Masthead or local use.TYPE 9006 Freq: 5-250MHz. B/W up to 40% of CF. Gain 10-40dBvariable. 50 ohms. NF 0.6dB. £105TYPE 9004 Freq: 250-1000MHz. B/W up to 10% of CF. NF 0.6dB.Gain 25dB. 50 ohms. £135TYPE 9304 Freq: 1-2GHz. B/W up to 10% of CF. NF 0.4dB Gain20d13.50 ohms. £185TYPE 9035 Transient protected mains power supply for aboveamplifiers. £58TYPE 9010 Masthead weatherproof unit for above amplifiers. £16

PHASE LOCK FREQUENCY CONVERTERSTYPE 9315 Down converter. lip frequencies 250MHz-2GHz. 0/p frequencies20MHz-1 GHz. B/W up to 10MHz. NF 0.7 dB. Gain 30dB variable. £350

TYPE 9316 Up/down converter. I/p & o/p frequencies 20MHz-2GHz. B/W upto 100MHz NF 0.7dB. Gain 40dB variable. £550

TYPE 9115A Up/down converter. I/p & o/p frequencies 20MHz to 1GHz. B/Wup to 100MHz. NF 0.7dB. Gain 60dB variable. 0/p up to 10mW + 10dBm.AGC. £650

VOLTAGE TUNABLE DOWN CONVERTERTYPE 93171/p will tune 30% of CF specified in the range 300MHz-2GHz. 0/p70MHz NF 0.6B. Gain 60dB. 0/p up to 10mW +10dBm. AGC. £950

PHASE LOCK SIGNAL SOURCES 20-2000MHzTYPE 8034 Freq. as specified in the range 20-250MHz 0/p 10mW. £194

TYPE 9036 Freq. as specified in the range 250-1000MHz. 0/p 10mW. £291

TYPE 9038 Freq. as specified in the range 1-2GHz. 0/p 10mW. £350

TYPE 9282 FM up to ±75KHz max. Freq. as specified in the range 30-2000MHz. 0/p 10mW. £378

WIDEBAND AMPLIFIERSTYPE 9301 100KHz-500MHz. NF 2dB at 500MHz. Gain 30dB. Output12,5dBm, 18mW. 50 ohms. £165

TYPE 9302 10MHz-1GHz. NF 2dB at 500MHz. Gain 30dB. Output 12.5dBm,18mW. 50 ohms. £165

TYPE 9008 Gasfet. 10iv1Hz-2GHz NF 2.5dB at 1GHz. Gain 10dB. Output18dBm. 65mW. 50 ohms. £165

TYPE 9009 Gasfet. 10MHz-2GHz. NF 3.8dB at 1GHz. Gain 20dB. Output20dBm, 100mW. 50 ohms. . £185

WIDEBAND LINEAR POWER AMPLIFIERSTYPE 9246 1 watt output. 100KHz-175MHz. 13dB gain. £192

TYPE 9036 1 watt output. 10MHz-1GHz. 15dB gain. £312

TYPE 9247 4 watt output. 1-50MHz. 13dB gain. .. . £215

TYPE 9051 4 watt output. 20-200MHz. 13dB gain. £215

TYPE 9176 4 watts output. 1-50MHz. 26dB gain. £345


TYPE 9178 10 watts output. 1-50MHz. 13dB gain £304







UHF LINEAR POWER AMPLIFIERSTuned to your specified frequency in the range 250-470 MHzTYPE 9123 500mW input, 5 watts output. £350

TYPE 9124 2-3 watts input, 25 watts output £510

TYPE 9126 8 watts input, 50 watts output £1495

Prices exclude p&p charges and VAT

RESEARCH COMMUNICATIONS LTDUnit 1, Aerodrome Industrial Complex, Aerodrome Road, Hawkinge, Folkestone, Kent CT18 7AG

Tel: 0303 893631 Fax: 0303 893838

96



RESEARCH NOTES

High -density storage gets chrome finishC dentists working at the US NationalJInstitute of Standards and Technology(Nisi) have glimpsed a future where veryhigh density storage devices could be builtaround individual atoms of chromium.

The Nist team has used a laser to depositneat rows of individual chromium atoms ona silicon substrate. But the significantachievement is that the rows are a mere65nm wide - considerably smaller than thesmallest structure that can be created byconventional lithographic techniques.

Atom optics, as the technology is called,involves using a split laser beam to create astable interference pattern -a standing wave- just above a silicon substrate. The patternof standing waves consists essentially ofalternating light and dark bands, and into itis fired a beam of chromium atoms, allinside an evacuated chamber.

The light and dark bands behave like anarray of atomic lenses, focusing the chromi-um atoms and depositing an identical patternon the surface of the silicon. Chromiumatoms are deposited in the areas where thelight intensity is lowest.

According to the Nist researchers(Science, Vol 262, 877) the laser fieldsinfluence atomic trajectories by causingthem to absorb and re -radiate photons. They

also create a dipole force proportional to theintensity gradient in the oscillating electricfield of the laser. Both effects are at theirstrongest when the laser frequency coincideswith the atom's natural resonance.

AT&T Bell Laboratories first used thetechnique to deposit patterns of sodiumatoms. Bul sodium is not stable in air andthe resulting structure had poor life.

The Nist creation, being made of chromi-um, is more permanent. Next stage, accord-ing to Nist's Bob Celotta, is to create a two-dimensional optical standing wave thatwould allow deposition, not just of lines, butof dots. Careful movement of the substratewould then permit successive dots no biggerthan a few atoms in size to be deposited nextto each other in arbitrary patterns.

The ability to create such structures withdetails as small as 5nm (about 15 atoms)would clearly open up a whole new area ofnanotechnology.

Practical problems are still immense.Quantum effects, for example, play a bigpart and the team is currently investigatinghow thin a piece of chromium 'wire' can bebefore it ceases to behave electrically like apiece of wire.

But the choice of chromium will allow theresearchers to investigate more than mere

Hole story behind superfast p-mosfetCornell University electrical engineers

have fabricated a p -channel mosfet thatovercomes one of the seemingly -inherentdisadvantages of such devices - the poormobility of holes compared to electrons.

In conventional silicon microelectronics,holes travel three times slower thanelectrons, leading to the disparity inperformance between complementary n- andp -channel devices. But Cornell AssistantProfessor Yosef Shacham-Diamand andPhD student Kaushik Bhaumik havedeveloped a new type of p -channel devicewith a l0nm (50 atom diameters) layer ofsilicon -germanium. This layer cracks thespeed problem by creating a quantum wellthat captures holes, providing them with asort of conduction 'fast lane'. Shacham-Diamand says that, for an equivalentterminal voltage, the new device passes 40%more current (about 5mA) and switches 40%faster than a typical p -channel device.

"Now we have a p -channel device that'sjust as fast as an n -channel device," he says.

Electron -beam -lithography capabilities ofthe National Nanofabrication Facility atCornell were used to define the gate lengthof the transistor to less than 0.21.im. At thissort of gate dimension, the holes in the SiGelayer travel at speeds exceeding that of holes

in silicon, an enhancement reflected in theoverall performance of the device.

Another benefit is a possible reduction inthe operating voltage, in this case to I.5Vrather than the more common 3.3V. Theresultant increased chip density makesgreater complexity possible as well as fasterspeed. Despite the improvements, no special

Thin wet oxide

Ti silicide con t

111 AII111Laa

z -

Laser

Deposited linesStanding wave

Nanometre-wide lines of chromium aredeposited by a standing -wave laser field thatforms cylindrical lenses. These focus thechromium atoms into strips that could be thebasis of very high density storage devices.

electrical conductivity. Celotta points outthat chromium dioxide has magnetic proper-ties that are already widely exploited inmagnetic data storage. Given the ability todeposit the material in well -ordered arrays atthe atomic level, the prospects of being ableto develop advanced new storage technolo-gies are considerable.

Celotta says, "We are going to try to makemagnetic media with this material and thatshould lead to high density magnetic stor-age... Eventually we will also be pursuingother materials, such as silicon and galliumarsenide semiconductors".

fabrication techniques are needed and thereare no extra processing steps.

On test, the new p -channel mosfet wasclocked at 35GHz, substantially faster thanthe best that can be achieved with standardp -channel devices (about 10GHz), and evenmarginally faster than the equivalent n -channel mosfet (32GH7).

+ poly gate

'61.::::::=6:::.

'n type Silico4 A. ... . N4ft444,i,E*WOW.

p+ Source/Drain (LDD)

Strained SiGe quantum well

In complementary mosfet ICs, p -channel devices perform less well than their n -channelcounterparts due to low hole mobility. This new p -channel structure could lead to fastercomplementary mosfet ICs with the bonus of lower operating voltage and higher density.


RESEARCH NOTES

Jupiter's big bang could shake scientific worldThe scientific community is expectantlygrabbing its seats to observe - albeit

indirectly - one of the biggest collisions inthe Solar System in recent times. Theoccasion will be when over twentyfragments of a broken comet, on collisioncourse for Jupiter, hit the giant planet at aspeed of 60km/s in July. Unfortunately,Comet Shoemaker -Levy 9 will hit the sideof Jupiter facing away from the Earth. Evenso, the International Astronomical Unionsays it expects the comet to undertake a finaldramatic act of suicide in which thefragments will hit the surface of Jupiter witha force equivalent to a 100Gt nuclearweapon. The impact would be of the sameorder as that thought to have occurred 65million years ago on Earth and which mayhave led to the extinction of the dinosaurs.

Shoemaker -Levy 9, named after itsdiscoverers, is one of the strangest objects inthe solar system. Instead of being a singlebody, it looks more like a string of pearls inthe sky. Pictures taken by the Hubble spacetelescope show fragments of up to 5km indiameter, chasing each other in a processiontens of thousands of kilometres long.

The orbits of Jupiter and Shoemaker -Levyhad been getting progressively closer forsome time. But the history of the bizarremulti -comet actually goes back no furtherthan July 1992 when, as a normal comet, italmost crashed into Jupiter. Instead thepowerful gravity of Jupiter simply wrenchedthe comet into tiny pieces.

600,000 MILES

The now -awaited collision is so importantthat, in spite of the fact that it is happeningon the dark side of the planet, astronomersare thinking up ingenious ways to observewhat happens.

One idea is to look at Jupiter's variousmoons to see if they reflect any light fromwhat is bound to be a firework display onthe grand scale. Another idea is to look forrefracted light around the rim of the planet.

Jupiter rotates very fast on its axis soastronomers will not have to wait more thanan hour or two for the scars on the planet'ssurface to swing round into view. Butobservers hope to be able to see the impactdirectly - using Voyager 2. The spacecraft istoo far beyond Jupiter to resolve a clearimage, but it should still be able to measurethe intensity of the flashes of light.

The other hope is that Galileo, still on itsway to Jupiter, might able to catch anoblique view of what's going on behind.

Precisely what will happen when all thefragments of Shoemaker -Levy hit the giantred planet remains a matter of some debate.It is expected to be a bright and spectacularevent, with huge holes punched in Jupiter'satmosphere and with gigantic shock wavesreverberating around the planet. Researchersat Sandia National Laboratories inAlbuquerque are currently attempting to

'String of Pearls' comet on collisioncourse with Jupiter. Left is taken fromEarth, centre and right from Hubble.

100,000 MILES

gain a better understanding of thiscataclysmic event by using a supercomputersystem originally developed to modelnuclear weapon blasts.

Sandia's Mark Boslough says that theJovian collision will have certain differencesrelative to what happened here on earth 65million years ago. On earth, the impactwould have caused an instant atmosphericpressure rise, with all the force of a massivenuclear weapon. But because Jupiter doesnot have a solid surface, the impact mayhave different characteristics. Sandia'ssupercomputer simulation predicts that whenShoemaker -Levy 9 enters the atmosphere ofJupiter, it will at first slice throughunhindered. After that, the pressure willbuild up gradually until the comet piecesbreak up further.

At this point, each piece will have lost 2%of its kinetic energy. The remaining 98%will be carried beneath Jupiter's cloudswhere it will be explosively released.

Still to be worked out on this model iswhether the final big bang will result in agiant mushroom cloud, another Great RedSpot or nothing at all.

The answer, for which we may need towait until July, is more than just a matter ofcuriosity. Astronomers believe that impactsof this sort are the means by which theplanets were created in the first place. So theway in which Shoemaker -Levy 9 commitssuicide could answer some fundamentalquestions about our own history.

40,000 MILES


SMALL SELECTION ONLY LISTED - EXPORT TRADE AND QUANTITY DISCOUNTS -RING US FOR YOUR REQUIREMENTS WHICH MAY BE IN STOCK

Marconi TF2008 - AM -FM signal generator - Also sweeper - 10Kcis - 510Mcis - from £350 B&K 4812 calibrator head.tested to £500 as new with manual - probe kit in wooden carrying box. Famell power unit H60/50 - £400 tested.HP Frequency comb generator type 8406A - £400. H.P. FX doubler 938A or 940A -£300.HP Sampling Voltmeter (Broadband) type 3406A - £200. Racal/Dana 9300 RMS voltmeter - £250.HP Vector Voltmeter type 8405A - £400 to £600 - old or new colour. H.P. sweeper plug -Ins - 86240A - 2-8.4GHz- 86260A - 12.4-18GHz - 86260AH03 - 10 -HP Synthesiser/signal generator type 8672A -2 to 18GHz £4000. 15GHz - 86290B -2-18.6GHz. 86245A 5.9-12.4GHz.HP Oscillographic recorder type 7404A -4 track - £350. TelequIpment CT71 curve tracer - £200.HP Plotter type 9872B -4 pen - £300. H.P. 461A amplifier - 1kc-150Mc/s - old colour -£100.HP Sweep Oscillators type 8690 A & B + plug -ins from 10Mc/s to 18GHz also 18-40GHz. P.O.R. M.P. 8750A storage normalizer.HP Network Analyser type 8407A + 8412A + 8601A - 100KcJs -110Mc/s- £500-£1000. Tektronix oscilloscopes type 2215A - 60Mcis - c/w book & probe - £400.HP Down Converter type 11710B -.01-11Mc/s - £450. Tektronix monitor type 604 - £100.HP Pulse Modulator type 11720A -2-18GHz - £1000. Marconi TF2330 or TF2330A wave analysers- £100-£150.HP Modulator type 8403A - £100-£200. HP5006A Signature Analyser £250 + book.HP Pin Modulators for above -many different frequencies - £150. HP10783A numeric display. £150.HP Counter type 5342A - 18GHz - LED readout -£1500. HP 3763A error detector. £250.HP Signal Generator type 86406- Opt001 + 003 - .5-512Mc/s AM/FM - £1000. Racal/Dana signal generator 9082 - 1.5-520Mcls - £800.HP Amplifier type 8447A - .1-400Mc/s £200 - HP8447F .1-1300Mc/s £400. Racal/Dana signal generator 9082H -1.5-520Mc/s - £900.HP Frequency Counter type 5340A - 18GHz £1000 - rear output £800. Claude Lyons Compuline - line condition monitor - in case - LMP1 + LCM1 £500.HP 8410 -A -B -C Network Analyser 110Mc/s to 12GHz or 18GHz - plus most other units and Efratom Atomic FX standard FRT FRK .' 1 5-10Mc/s. £3K tested.displays used in this set-up -8411A -8412-8413-8414-8418-8740-8741-8742-8743 Racal 4D recorder - £350 - £450 in carrying tag as new.- 8746 - 8650. From £1000. HP8350A sweep oscillator mainframe + HP1 /869A RF PI adaptor -£1500.HP Signal Generator type 8660C - .1-2600Mc/s. AM/FM - £3000. 1300Mcis £2000. Aiitech - precision automatic noise figure indicator type 75 - £250.HP Signal Generator type 8656A -0.1-990Mc/s. AM/FM - £2000. Adret FX synthesizer 2230A- 1Mo-is. £250.HP 8699B Sweep Pi -0.1-4GHz £750 - HP8690B Mainframe £250. Tektronix -7S12 -7514-7711-7511-51-552-S53.Racal/Dana 9301A-9302 RF Millivoltmeter - 1.5-2GHz - £250-£400. Rotek 610 AC/DC calibrator. £2K + book.Racal/Dana Counters 9915M -9916 -9917 - 9921 - £15010 £450. Fitted FX standards. Marconi TF2512 RF power meter - 10 or 30 watts - 50 ohms - £80.Racal/Dana Modulation Meter type 9009 - 8Mc/s - 1.5GHz - £250. Marconi multiplex tester type 2830.Racal - SG Brown Comprehensive Headset Tester (with artificial head) Z1A200/1 -£350. Marconi digital simulator type 2828A.Marconi AF Power Meter type 893B - £200. Marconi channel access switch type 2831.Marconi RCL Bridge type TF2700 -£150. Marconi automatic distortion meter type TF2:137A- £150.Marconl/Saunders Signal Sources type - 60588 - 6070A - 6055B - 6059A - 60578 - 6056 - Marconi mod meters type TF2304 - £250.£250-£350. 400Mc/s to 18GHz. HP 5240A counter - 10Hz to 12.4GHz - £400.Marconi TF1245 Circuit magnification meter + 1246 & 1247 Oscillators -£100-£300. HP 3763A error detector.Marconi microwave 6600A sweep osc., mainframe with 6650 PI - 18-26.5GHz or 6651 PI - 26.5- HP 8016A word generator.40GHz - £1000 or PI only £600. HP 489A micro -wave amp - 1-2GHz.Marconi distortion meter type TF2331 -£150, TF2331A - £200. HP 8565A spectrum analyser -.01-22GHz - £4k.Microwave Systems MOS/3600 Microwave frequency stabilizer - 1GHz to 40GHz £tk. HP 5065A rubidium vapour FX standard - £5e.Tektronix Plug -ins 7A13- 7A14 - 7A18 - 7A24 - 7A26 -7A11 -7M11 -7511 - 7DI0 -7512 - Fluke 893A differential meters - MO ea.St - S2 - S6 - S52 - PG506 - SC504 - SG502 - SG503 - SG504 - DC503 - DC508 - DD501 - Systron Donner counter type 60548- 20Mc's-24GHz - LED readout -£1k.VVR501 - DM501A - FG501A - TG501 - PG502 - DC505A - FG504 - P.O.R. Takeda Riken TR4120 tracking scope + TR' 604P digital memory.Aiitech Stoddart receiver type 17/27A - .01-32Mc/s - £2500. EG&G Parc model 4001 indicator + 4203 signal averager Pl.Aiitech Stoddart receiver type 37/57- 30-1000Mc/s - £2500. Systron Donner 6120 counter/timer A +B +0 inputs - 18GHz -£1k.Aiitech Stoddart receiver type NM65T - 1 to 10GHz -£1500. Racal/Dana 9083 signal source - two tone - £250.Gould J3B Test oscillator + manual - £200. Systron Donner signal generator 1702- synthesized to 1GHz - AM/FM.Infra -red Binoculars in fibre -glass carrying case - tested - £100. Infra -red AFV sights £100. Systron Donner microwave counter 6057- 18GHz- Nixey tube - £600.ACL Field intensity meter receiver type SR - 209 - 6. Plugs -ins from SMc/s to 4GHz - P.O.R. Racal/Dana synthesized signal generator 90E11 - 520Mc/s -AM-FM. £600.Tektronix 491 spectrum analyser - 1.5GHz-40GHz - as new - £1000 or 10Mc/s 40GHz Fame!! SSG520 synthesized signal generator - 520Mc/s - £500.Tektronix Mainframes - 7603 - 7623A - 7633 - 7704A - 7844 - 7904 - TM501 - TM503 - Farnell TTS520 test set - £500 - both £900.TM506 - 7904 - 7834 - 7104. Tektronix plug -ins - AM503 - PG501 - PG508 - PS503A.Knott Polyskanner WM1001 + WM5001 + WM3002 + WM4001 - £500. Tektronix TM515 mainframe + TM5006 mainframe.Aiitech 136 Precision test RX + 13505 head 2 - 4GHz - £350. Cole power line monitor T1085 - £250.SE Lab Eight Four - FM 4 Channel recorder - £200. Claude Lyons LCM1P line condition monitor - £250.Alltech 757 Spectrum Analyser - 001 22GHz - Digital Storage + Readout - £3000. Rhodes & Schwarz power signal generator SLRD-280 - 2750Mc1s. £250-L600.Dranetz 606 Power line disturbance analyser - £250. Rhodes & Schwarz vector analyser - ZPV4 El +E3 tuners - .3-2000Mds.Precision Aneroid barometers- 900-1050Mb - mechanical digit readout with electronic indicator Bell & Howell TMA3000 tape motion analyser - £250.- battery powered. Housed in polished wood carrying box - tested - £100-1200-£250. 1, 2 or 3. Ball Efratom PTB-100 rubidium standard mounted in Tek Pl.,HP1417 SPECTRUM ANALYSERS- ALL NEW COLOURS Ball Efratom rubidium standard PT2568-FRKL.TESTED WITH OPERATING MANUAL Trend Data tester type 100 - £150.HP141T -, 8552A or B IF -8553B RF -lkHz-110Mc/s-A IF - £1300 or B IF - £1400. Farnell electronic load type RB1030-35.HP141T+ 8552A or B IF -8554B RF -100kHz-1250Mc/s-A IF -£1400 or B IF -£1500. Fairchild interference analyser model EMC 25 - 14kcis-1GHz.HP141T+ 8552A or B IF -8555A RF - 10Mc/s-18GHz-A IF - £2400 or B IF - £2500. Fluke 1720A instrument controller + keyboard.HP141T+8552A or B IF -8556A RF - 20Hz-300kHz-A IF -A IF - £1200 or B IF -£1300. Marconi 2442 -microwave counter - 26.5GHz - £1500.HP8443A tracking generator/counter - 100kHz-110Mc/s - £500. Racal/Dana counters - 9904 - 9905 - 9906 -9915 -9916 - 9917 - 9921 - SOMcis - 3GHz -HP8445B tracking pre -selector DC-18GHz - £750. £100-£450 - all fitted with FX standards.HP ANZ UNITS AVAILABLE SEPARATELY - NEW COLOURS -TESTED. B&K 7003 tape recorder - E300.HP1417 mainframe - £550 - 8552A IF - £450 - 8552B IF - £550 - 8553B RF -1kHz-110Mc/s - B&K 2425 voltmeter - £150.£550 - 8554B-RF - 100kHz-1250Mcis - £650 - 8555A-RF - 10Mc/s-18GHz -£1550. B&K 4921+4149 outdoor microphone.HP 3580A LF-spectrum analyser - 5kHz to 50kHz - LED readout - digital storage -£1600 with Wlttron sweeper mainframe 610D - £500.instruction manual - internal rechargeable battery. HP3200B VHF oscillator - 10-500Mc/s - £200.Tektronix 7D20 plug-in 2 -channel programmable digitizer - 70 Mc/s - for 7000 mainframes - HP3747A selective level measuring set.£500- manual - £50. HP3586A selective level meter.Datron 1065 Auto Cal digital multimeter with instruction manual - £500. HP5345A electronic counter.Racal MA 259 FX standard. Output 100kc/s-1Mc/s-5Mc/s - internal NiCad battery - £150. HP4815A RF vector Impedance meter c/w probe. £500-£600.Aerial array on metal plate 9"x 9" containing 4 aerials plus Narda detector -.100-11 GHz. Using Marconi TF2092 noise receiver. A, B or C plus filters.N type and SMA plugs & sockets - ex eqpt -£100. Marconi TF2091 noise generator. A, B or C plus filters.EIP 451 microwave pulse counter 18GHz - £1000. Tektronix oscilloscope 485 - 350Mc/s - £500.Marconi RF Power Amplifier TF2175 - 1.5Mc/s to 520Mc/s with book -£100. HP180TR, HP1827 mainframes £300-£500.Marconi 6155A Signal Source -1 to 2 GHz - LED readout - £600. Bell & Howell CSM2000B recorders.Schiumberger 2741 Programmable Microwave Counter - 10Hz to 7.1GHz - £750. HP5345A automatic frequency convertor - .015-4GHz.Schlumberger 2720 Programmable Universal Counter 0 to 1250Mc/s - £600. Fluke 8506A thermal RMS digital multimeter.HP 2225CR Thinkjet Printer - £100. HP3581A wave analyser.TEK 576 Calibration Fixture - 067-0597-99 - £250. Philips panoramic receiver type PM7800 - 1 to 20GHz.HP 8006A Word Generator -E150. Marconi 6700A sweep oscillator+ 6730A -1 to 2GHz.HP 1645A Data Error Analyser - E150. Wiltron scaler network analyser 560+ 3 heads. £1k.Texscan Rotary Attenuators - BNC/SMA 0-10-60-100C:43S - £50-£150. R&S signal generator SMS -0.4-1040Mc/s -E1500.HP 809C Slotted Line Carriages - various frequencies to 18GHZ -£100 to £300. HP8558B spectrum ANZ PI -.1-1500Mes - o/c -C1000. N/C -C1500 -To fit HP180 seriesHP 532-536-537 Frequency Meters - various frequencies - £150-£250. mainframe available -E100 to £500.Barr & Stroud variable filter EF3 0.1Hz-100kc/s + high pass + low pass - £150. HP8505A network ANZ + 8503A S parameter test set + 8501A normalizer - £4k.S.E. Lab SM215 Mk11 transfer standard voltmeter - 1000 volts. HP8505A network ANZ + 8502A test set - £3k.Aiitech Stoddart P7 programmer - £200. Racal/Dana 9087 signal generator - 1300Mc/s - £2k.H.P. 6941B multiprogrammer extender. £100. Racal/Dana VLF frequency standard equipment. Tracor receiver type 900A + differenceFluke Y2000 RTD selector + Fluke 1120A IEEE -488 -translator - Fluke 2180 RTD digital meter type 527E+ rubidium standard type 9475 - E2750.thermometer + 9 probes. £350 all three items. Marconi 6960-6960A power meters with 6910 heads - 10Mc/s - 20GHz or 6912 - 30kHz-H.P. 6181 DC current source. £150. 4.2GHz -£800-£1000.H.P. 59501A - HP-IB isolated D/A/power supply programmer. HP8444A-HP8444A opt 59 tracking generator Elk-C2k.H.P. 3438A digital multimeter. B&K dual recorder type 2308.H.P. 6177C DC current source. £150. HP8755A scaler ANZ with heads £1k.H.P. 6207B DC power supply. Tektronix 475 - 200Mc/S oscilloscopes - £350 less attachments to £500 c/w manual, probes etc.H.P. 741B AC/DC differential voltmeter standard (old colour) £100. HP signal generators type 626 - 628 - frequency lOGHz-21GHz.H.P. 6209B DC power unit. HP 432A -435A or B -436A - power meters+ powerheads - 10Mc/s-40GHz - £200-£280.Fluke 80 high voltage divider. HP3730B down convertor - E200.Fluke 4310 high voltage DC supply. Bradley oscilloscope calibrator type 192 - £600.Tektronix M2 gated delay calibration fixture. 067-0712-00. Spectrascope SD330A LF realime ANZ- 20Hz-50kHz - LED readout - tested - £500.Tektronix precision DC divider calibration fixture. 067-0503-00 HP8620A or 86200 sweep generators -1250 to £1k with IEEE.Tektronix overdrive recovery calibration fixture. 067-0608-00. Barr & Stroud variable filter EF3 0.1Hz-100kc/s+ high pass+ low pass -£150.Avo VCM163 valve tester+ book £300. Tektronix 7L12 analyser -.1Mc/s-1.8GHz -£1500 - 7L14 ANZ - £2k.H.P. 5011T logic trouble shooting kit. £150. Marconi TF2370 spectrum ANZ -110Mcds - £1200-£2k.Marconi TF2163S attenuator - 1GHz. £200. Marconi TF2370 spectrum ANZ+TK2373 FX extender 1250Mc/s+trk gen - £2.5k -£3k.PPM 8000 programmable scanner. Racal receivers -RA17L-RA1217-RA1218-RA1772-RA1792 - P.O.R.Fluke 730A DC transfer standard. Systron Donner microwave counter 6057 - 18GHz - nixey tube - £600.B&K 4815 calibrator head. HP8614A signal gen 800Mc/s-2.4GHz old colour £200, new colour £400.

HP8616A signal gen 1 8GHz-4.5GHz old colour £200, new colour £400.

ITEMS BOUGHT FROM NM GOVERNMENT BEING SURPLUS. PRICE IS EX WORKS. S.A.E. FOR ENDUIRIES. PHONE FOR APPOINTMENT OR FOR DEMONSTRATION OF ANY ITEMS. AVAILABILITY OR PRICE CHANGE. VAT AND CARR., EXTRA.

ITEMS MARKED TESTED HAVE 30 -DAY WARRANTY. WANTED: TEST EOPT - VALVES - PLUGS & SOCKETS - SYNCROS - TRANSMITTING & RECEIVING EQPT. ETC.

Johns Radio, Whitehall Works, 84 Whitehall Road East, Birkenshaw, Bradford BD11 2ER. Tel. No. (0274) 684007. Fax 651160.


February 1994 ELECTRONICS WORLD+WIRELESS WORLD 99

AUDIO

13AdingCoupled -cavityloudspeakers have thecombined advantages ofextended bass responseand shallow If roll -off.lan Gosling highlightsthe design process witha worked sub -wooferexample.

n recent years manufacturers have fol-lowed a trend towards smaller speaker enclo-sures. Sparsely furnished rooms exhibit unde-sirable room resonance at around 30Hz, andsmall speakers are more popular for aestheticreasons. As a result, few reasonably pricedsystems now offer a -3dB point much lowerthan 70Hz.

Bass response has suffered. This is a pity,since the bottom octaves of piano and bassguitar both extend down to 40Hz, and percus-sion much lower. Many commercial systemsuse vented boxes to reduce footprint, but withthe penalty of steep response roll -off. Ventingalso degrades transient response - an impor-tant factor in perceived sound quality.

Compact discs can reproduce the full audiospectrum from 2Hz to 20kHz. I thereforelooked for a suitable sub -woofer design toextend my existing system. Its specificationwas a frequency response flat down to 20Hz,with a gradual roll -off of 12dB/octave or better.

I chose 80Hz as the subwoofer crossoverfrequency. There are a few choices of enclo-

I Ou ELECTRONICS WORLD + WIRELESS WORLD February 1994

I AulY11,11

AUDIO

sure, each with their own merits. I chose thecoupled -cavity enclosure. It is smaller than atransmission line and much smaller than ahorn. In addition, radiation is only emittedfrom the port, which is quite small. Thisreduces the area of grille cloth needed, makingthe unit surprisingly inconspicuous in adomestic environment.

Since the high frequency response of thecavity -based subwoofer rolls off naturally, acrossover network is not needed. If you wishnot to enter into filter theory, a rough designfor a simple closed box can be completed in afew minutes using a calculator. For more com-plex enclosure types however, a first cutdesign procedure can give disappointingresults. It is a good idea to model the fre-quency response on a computer before startingto cut timber.

The enclosure can be modelled as an acous-tic circuit on a Spice type circuit simulator.Prices of such software nowadays can bemuch less than the cost of speaker materials.

Acoustic circuitsJust as voltage V and current I describe anelectrical signal, so a sound wave can bedescribed by air pressure p in N/m2 and therate of movement of air or volume velocity 1,in m/s, Fig. 1. Equivalent electrical compo-nents can represent acoustical loudspeakercomponents.

Consider first a sealed box with an air inlet,Fig. 2. When air is pumped in, the pressurerises. This is analogous to an electrical capac-itor, where injecting charge causes the poten-tial difference to rise. The acoustic equivalentof capacitance is the acoustic compliance CAmeasured in m/N.

A similar argument applies to air beingpumped into the space near an elastic mechan-ical part such as the cone suspension. Whendeflected from it rest position, it exerts a pres-sure on the air through the attached cone. Theacoustic compliance is proportional to themechanical compliance or spring constant.

Mass of the drive unit cone and the mass ofair trapped in the port tube also have acousticequivalents. A mass moving at the samevelocity as the air behind it is in equilibriumand no forces act. If the air velocity changessuddenly, the mass cannot react instanta-neously to move with it, so the air starts topress on the mass. This is analogous to changeof current through an inductor, which causes aback emf.

The equivalent of inductance is acousticmass MA measured in kg/m4. Radiation ofsound energy is analogous to electrical powerdissipation. As a result, the circuit componentis an acoustic radiation resistance. There isalso a reactance due to the mass of air trappedjust in front of the radiating surface.

Finally, the electrical part of the circuit mustalso be modelled. An electrical resistance REappears through the transformer action of thevoice coil motor as an acoustic resistance

(BI)2RA -- Si, R

where B1 is magnetic field in the voice coilmultiplied by the coiled length of wire and REis the sum of amplifier output impedance,speaker cable resistance and voice coil resis-

V V V

tance. Typically, RE is around 852. Sensiblychosen speaker cable has a negligible contri-bution to this figure - another nail in the cof-fin for fancy cables.

Enclosure design process Fg. 3 shows theenclosure configuration. Two drive units areused to reduce the large cone excursion nec-

p

high p high p

V V _

low p

Fig. 1. Voltage and current in a travelling wave on an electrical two -wire transmission linecompared to pressure and velocity in a sound wave. The air has maximum and minimumdensity at the places marked + and - respectively.

t

aDI

(NOP

CA = wipe

CA =

MA = MiSD2

M MOO MA = (t+0.6a0p/S,

ZAR = co2p/47cC

+ 0.23j)/a,

Fig. 2. Top to bottom, electrical components analogous to a sealed box of volumeVB, a cone suspension of spring constant CM, a cone of mass m,a port tube of volume SPxt and a radiating surface of radius ao. Parameter p is thedensity of air, c the speed of sound and So cone surface area.


AUDIO

essary at low frequencies.Behind each is a sealed box of volume VB.

One drive unit may be mounted back to frontto reduce second harmonic distortion. In thiscase its electrical connections must also bereversed.

In acoustic circuit Fig. 4, R1.2 are the voicecoil resistances, R3A the energy absorbed inthe cone suspension, L12 the cone masses andC1.2 the suspension compliances. C3.4 repre-sent the compliance of air in boxes VB. Sinceall these components move together, they dis-place the same amount of air, so must sharethe same volume velocity r. This is equivalentto electrical components that all carry thesame current. They are therefore connected inseries.

The cavity of volume Vc (C5) can be regard-ed as a sealed box into which air can flow

from the drive units and from the port tube L3.Since these volumes of air add, there is a vol-ume summing junction at one end of C5,equivalent to a current summing junction.

For a closed box, the acoustic circuit has ahigh-pass resonance below which there is nouseful sound output. Inspection of this circuitshows that there is a band-pass resonance.Since this lies at the centre of the operatingfrequency band, the lower 3dB point can he atleast an octave below that obtainable withsimpler enclosure types.

The schematic can be entered on a PC -basedcircuit simulator and output sound pressurelevel plotted as the voltage across the resistivepart of the radiation impedance Z1.

Alternatively, a short program can be writtento perform the circuit calculations and displaythe results. Enclosure and port dimensions are

then adjusted to obtain the required response.For a speaker response extending down to

20Hz, the enclosure resonant frequency shouldbe about 40Hz and the free air resonance ofthe drive units, J: should be less than 30Hz.Manufacturers provide drive unit parameters,but they also can be measured as shown in thepanel.

I used the KEF B1398. which has low dis-tortion and a free air resonance of 29.5Hz. Fora design procedure not involving filter theory,initially the box volume VB can be made infi-nite. Parameter Vc and the acoustic mass ofthe port are chosen to give the required centrefrequency. Then VB is reduced to a sensiblesize and final adjustments made to the band-width and response shape adjusted.To meet my specification, VB should be0.08m3, Vc is 0.056m3 and port cross-section

Bass speaker enclosures

Loudspeaker drive units were first designed for use with aninfinite baffle. A closed back to the enclosure degraded theperformance rather than enhancing it.

More recent air suspension drivers are designed to workwith an enclosure. Compressing the air in the sealed boxcreates pressure on the cone, and this can be employed toprovide the restoring force for the cone as opposed to thetraditional stiff cone suspension. Sealed enclosures have ahigh-pass response with a gradual low frequency roll -off of12dB/octave.

Vented or reflex enclosures have an additional tubeterminated by a port which can radiate in addition to thedrive unit. This lowers the cut-off frequency of the responsesomewhat, but increases the roll -off slope to 18dB/octave.

In the transmission line enclosure, the port tube islengthened to be a half wavelength at the cut-off frequency.Sound then emerging from the port isin phase with that directly radiatedfrom the drive unit. This increases thesound output, extending the bassresponse. But the tube now becomes8.7m long at 20Hz. In addition, atfrequencies where the port is a wholenumber of wavelengths long,destructive interference occurs,

causing dips in the response.Fibre packing in the port tube can help by attenuating the

signal at mid bass frequencies and by reducing thewavelength by up to three times, as in the Bailey designs.Packing however causes pressure build-up behind the conewhich can spoil the frequency response.

Sound output at low frequencies can be improved bymaking the cone effectively larger by attaching it to a horn.Its mouth should ideally be at least one wavelength incircumference, i.e. 4.3m across for a flat response down to20Hz. The horn enclosure has a band-pass response with avery shallow If roll -off of 6dB/octave.Coupled -cavity enclosures2 have a cavity and a port in frontof the drive unit. This also results in a band-pass response.The lower 3dB point is well below that of a comparableclosed box while roll -off is unchanged.

(a)

51(b) (c) (d)

I

(e)

soundoutputdB spl

0-

10

coupled cavityTX line

closed boxvented boxhorn

10 20 50 100 200frequency Hz

500

Fig. 1. Speaker enclosure types.Diagram (a) is basic closed box, (b) isa vented box, (c) represents atransmission line, (d) a horn and (e) acoupled cavity.

Fig. 2. Comparison of frequencyresponses of different enclosures ofbroadly similar floor footprint. Thesecurves can vary considerably with theenclosure design and drive unitparameters.


AUDIO

Measuring driveunit parameters

If the drive unit parameters areunknown, the following prtocedurecan be used to obtain reasonablyaccurate values using readilyavailable instruments.

First measure area SD as the areaof the cone itself plus half thesurround. Measure across the frontof the chassis, not along the slope ofthe cone. Next apply a small weightto the cane. Suspension complianceCAS is then SD2 multiplied by theratio of cone deflection to the massapplied.

Now measure the DC voice coilresistance RE. Connect to a signalgenerator via a voltmeter andammeter and find the free airresonant frequency, is, wherecurrent is at its minimum. Divide theAC voltage by the current to find thetotal resistance of the drive unit,calling this RE+REc. Subtract RE toobtain REc.

It Bi is not known, measure thefrequencies f1 and f2 at which ACresistance is,

'AIRECRE + REC)

Calculate the electrical

QEc=f5412-f,),

and hence the component value

RLuiz

using Ri=1 /(21rfsQEcCAs)- CatzulateRAs=RTRE/REc. Then finally,

1

0 4611-S°MAD =47r2fs2CAs x

The second term allows for the massof air trapped next to the cone.

590

95

Fig. 3. Twin -unit coupled cavity enclosure. Circular holes are for the drive units. Sound outputcomes from the open end of the port Sp. A smaller model using only one drive unit may beconstructed by dividing the drawing in half along the centre line of the port tube. Thespeakers are mounted nose to tail.

drive units

R,

from ampO

R3 Li C,

boxVBC,

CAB

4

M, CA, CA, C,C,

cavtty

port

La

NA,

Z, soundZA,, output

Fig. 4. Acoustic circuit corresponding to Fig. 3. Since all the components move together, theydisplace the same amount of air and have the same velocity.

/////

Fig. 5. Resifts of thecomputer model.Cone velocity is low

0

dB //

Sound output

.'"------

at low frequencies,whereas in a ventedbox enclosure, thecone would bereaching the limits ofits excursion.

\\\

//

/1

I Co ne velocity4------

1 \ I

1 \ I-10 I 1

;

1 I

1 I

1 I

1 I

1

1

1

I

-20 I

0 50 100 150 Hz


AUDIO

0

dB

4510Hz

ry

frequency

90

180

1 kHz outer rad.=ampl. 1.4

Fig. 6 In a normal crossover network for loudspeakers, signal issplit into low -frequency and high -frequency components usingstandard filters, and the components are then recombined (solidline). The recombined signal has a peak at the crossoverfrequency, and there is a 180' phase shift across the band.

0

dB

-4510Hz frequency 1 kHz outer rad.=ampl. 1.4

Fig. 7. Using the high-pass filter described in the text, when low andhigh -frequency components are recombined, the resulting signal is flatacross -the band and has no phase shift.

4

180

I

90

Sp is 0.017m2 by length of 0.5m. This gives analmost ideal response, Fig. 5, with slightlyunder -damped low and high frequency roll -off.

Electronic filter designElectronic circuitry is required before thepower amplifiers to separate the electricaldrive for the sub -woofer from that for themidrange and treble. However, there can bepitfalls if this is done without care. In a loud-speaker crossover network, the electrical sig-nal is separated into low- and high -frequencycomponents using a low-pass and a high-passfilter. The two components are recombined bysummation in the listener's ear.

A simple circuit simulation using second -order filters with Q=0.7, Fig. 6, shows that,surprisingly, the recombined signal does nothave a flat frequency response. The low fre-quency component is also in antiphase to thehigh frequency one. This is usually correctedby reversing the connections to one speakerdrive unit - in which case the frequencyresponse of the recombined signal has a dip.

If the circuit is modified so that the high-pass filter is made from a duplicate low-passfilter plus a differential amplifier, Fig. 7 thenthe effect of the identical low-pass filters can-cels out exactly when the signals are recom-bined at the ear. The resulting frequencyresponse is perfectly flat and the phase of theresponse is 0° at all frequencies.

Figure 8 shows the filter circuitry for thecoupled cavity subwoofer. Low-pass filteraction is accomplished by the speaker acousticcircuit, so no high -power crossover network isrequired. The duplicate low-pass filter is aSallen and Key op -amp filter, U2. Subtractionis performed by U3: the output from U3 is thentaken to the midrange power amplifier.

Drive for the sub -woofer drive is obtainedby summing the stereo channels in U4A. Aphase shifter, comprising U4B and U5, may beincluded to correct for the room position of thesub -woofer. It comprises a very low -Q LCRall -pass filter using the gyrator U5B as theinductor, C6 and R28. It gives 180° adjustmentrange; the other 180° is covered by swappingover the speaker cables.

ConstructionChipboard or medium density fibreboard aresuitable materials for the enclosure. The twolarge unbraced panels should be 25mm thick,while the remainder is 18mm. KEF suggestsconnecting the drive units with a metal rod togive extra stiffness, but this requires custom -designed drive units. Here the port tube pro-vides the bracing.

Length t of the port tube (see Fig. 3)includes the length of the bend where it joinsthe cavity Vc -a component not included inthe acoustic circuit. Damping material shouldbe fixed to the inside of the panels to reducecabinet vibration and to eliminate high fre-quency cavity resonances, since there is nocrossover network. I used fibreboard pressuretreated with bitumen, generously glued allover. This was followed by a 50mm layer ofopen cell polyester foam cushion stuffing. Inpractice, this also favourably reduces the Q ofthe response shape.

The port tube should not be lined. The bot-tom face should not be glued, but screwed onto allow interior access, using a foam gasket.All other joints should be glued and screwedusing 25mm square battens, ensuring airtightjoints. A grille cloth may be added, but thematerial should offer minimum air resistanceas the air volume velocity in the port is high.

References1. Bailey, A.R., Non -ResonantLoudspeaker Enclosure Design, HighFidelity Designs, Wireless World, 1974,pp 87-90.2. Berriman, D., Sound Concepts ,

Electronics World and Wireless World,Sep. 1990, pp 774-779.

Further readingBeranek, L.L., Acoustics , AmericanInstitute of Physics, 1986.Small, R.H., Closed -Box LoudspeakerSystems, Journal of Audio EngineeringSoc., 2 parts, Dec. 1972, Jan/Feb. 1973.

Coupled -cavity design software

Turbo Pascal listings for plottingfrequency response and time -domainsquare -wave response of a twin -driver coupled -cavity speaker isavailable on disk. It can be obtainedby sending £10 plus vat toEW&WWs editorial offices at theaddress in the front of the magazine.This PC format software is not a fully -worked speaker design package butrather intended as a template for thecomputer -literate experimenter. Itcomprises three pascal files, eachover 5K. Documentation is limited toREM statements in the listings.


L in

R in

RS150k

01A

R1a

LM833

-vcc

+15V

ov

=15V 0

R6 150k

018

Rlb

R2a

R2b

150k

150k

100k

1%

1%

1%

Cla .7=7. Clb .7=-4 Clc

6n8

6n8

6n8

1%

1%

1%

VCC

47k

2%

C2a

4n7

1%

C2b

1n2

5%

R20

u2A

V 100k

1%

LM833

-VCC

LM833

_IL C9 47u

C10

25V T100n

-L

CliT

47u

C12

25V T100n

R3a

R3b

R4a

R4b

150k

150k

100k

1%

1%

1%

-1-

C a

13

-rC3b

T6

---

11k

1n8 T 6%8

1%

1%

VCC

47k

2%

C4a

4n7

1%

C4b

1n2

5%

R19

100k

1%

VCC

U3A

NE5532

818

-1,tC

C15

100k

1%

-

R15

110p

R24

028

100k

1%

R23

LM833

100k

1%

R22

100k

1% VCC

lin

lout

OV

filter

U3B

OV

NE5532

VCC

rlA

100k

VCC

R7 22k

0421

---:"..----

LM833

vcc

6Iv

R9

47k

1 -VCC

C5 IOp

10p

811

810

100kLOO

gain

VCC

U48

R21

10p

100k

1%

Rroc716c

VCC

USA

LM833 C8

-VCC

828

20k

1%

1%

LM833 Cli

-VCC

R17

100k

1%

I10p

IOp

10p

R12

R26

100kT,OC

R25

R14

2k4

100k

100k

C6

1%

1%

C7

ov ov

22n

5%

C14

220n 5%

R27

10p

2k4

-VCC

058

R13

100k.LOG

LM833

C

phase

shift

O

Oco

L midrange

out

R midrange out

subwoofer out

/

PROTEUSr Ii,; IrLnvf.D I T 1,301;.1 GE,' RA

4 SE-

(AMIE

ltt RE.

9

,t I

r 4 I:

. . .

I ' ".tt!irr: tr.:(1, ' t tr'innial Kt TA ..; Sitn MTN t trt REF t,'..."tot

IS. !

Hun..

vio 1441644 _

'P111117;616

cocenL,enElectronics

44 1 4,i! , k

,

O Easy to Use Graphical interface.Netlist, Parts List & ERC reports.

(i) Hierarchical Design.

O Extensive immponentimodel libraries.O Advanced Property Ma'agementO Seamless integration with simulation and

PCB design.

O Non -Linear & Linear Analogue Simulatiori.O Event driven Digital Simulation with

modelling language.

0 Partitioned simulation o large designswith multiple analogue & digital sections.

O Graphs d splayed directly on theschematic.

Multi-LayerMulti-Layer and SMT support.

Unlimited Design Capacity.

Full DRC and Connectivity Checking.

Advanced Multi -Strategy Autorouiing.

Output to )rinters, plotteis, Postscript,Gerber, DXF and DTP bitmaps.

Gerber View and Import capability.

Write, phone or fax for your free demodisk, or ask about our full evaluation kit.Tel: 0274 542868. Fax: 0274 481078.

14 Marriner's Drive, Bradford BD9 4JT.

Proteus software is for PC 386 compatibles and runs under MS-DOS. Prices start from £475 ex VAT; uli system costs £1495.Call for information about our budget, educational & Windows products. Al manufacturers' trademarks acknowledged.


AUDIO

It is now nearly seventy years since theintroduction of the moving coil speakerand small speaker systems with extended

bass response are still rare. In a way this is notsurprising. A thorough understanding of howspeaker systems operate in the bass region hasonly been available since the 1970s. Inaddition, the work of Thiele', extended andenhanced by Small'- and others, has onlyrecently been widely disseminated. This workforms the backbone of modern If speakerdesign.

An unmounted loudspeaker can be modelledby an electrical second order high pass filter.Typical response curves are shown in Fig. 1.Note that the response shape depends on the Qof the filter. Usually, the maximally flatButterworth filter is assumed to be the bestsince it combines maximum pass bandresponse flatness without peaking. Howeverthe best transient response is obtained with aQ of 0.5, although with the penalty of adrooping If response.

Filter action is a byproduct of the speaker'sfundamental resonance. This resonance occursbetween the mass of the cone and thecompliance of the cone surround. Responsecurve Q is mainly determined by electrical

I P Macaulay has thrown his subwoofer away,arguing that the best route to good low -

frequency response is to combine theadvantages of a small reflex enclosure withelectronic compensation.

mall boxdamping, or lack of it, imposed by the magnetassembly.

Antiphase bass radiated from the rear of thecone cancels radiation from the front. Sealingthe rear of the speaker in an enclosure solves

the problem but at the expense of increasingthe driver's resonant frequency. This is due toof compliance of the enclosed air effectivelystiffening the surround.

Using Thiele/Small theory, the response of

Fig. 1. Typical curvesfor a sealed

loudspeaker enclosure.Designers usually

strive fora Q of 0.7 asin curve B. Curve A is

fora Q of 2 whilecurve C is 0.5.

dB

+5

-10 -

-15

-20 -

-25 -

-30

-35

10

IR

100 1000Hz


AUDIO

dB

-5

-10

-15

-20

-25

-30

-35

10 100Hz

1000

Fig. 2. Sixth -ordercurve C is systemresponse producedby combiningfourth -order roll offof a reflex speakerwith second -orderelectronic low passfiltering.

such a system can be easily be calculated - atleast for the bass region. Three basicparameters of the driver are needed. First isthe bass resonant frequency, is., measured infree air and second Vas, the volume of airwhose compliance is equivalent to thecompliance of the speaker. Finally the parallelequivalent of the mechanical and electricalbass resonant Qs, or Qts, needs to be known.Nowadays any reputable driver manufacturershould be able to supply these details on theirproducts.

The other widely used method of mountinga speaker is to introduce a duct or port into theotherwise sealed enclosure - the bass reflexsystem - as detailed in the panel. By jugglingthe speaker parameters, enclosure volume andport tuning it is possible to devise a goodreflex system. But due to physical restraints,most affordable, reasonably -sized systems endup with a -3dB point somewhere between 50and 70Hz. Considering that most program

Reflex enclosure designRather than being airtight, the reflexenclosure has a duct or vent. Air

mass in the duct forms a resonantcircuit with the compliance of the air inthe cabinet -a form of the Helmholtzresonator.

At resonance, the reflex port invertsthe phase of the cone's rear radiation sothe duct radiates sound in phase withthe front of the speaker. This actionincreases efficiency at low frequencies.As an added bonus the speaker sees ahigh impedance at the resonantfrequency of the cabinet and coneexcursion is greatly reduced for a givenacoustic output. Furthermore theresonant frequency of the speakermounted in the cabinet is hardly raisedat all from its free air value.

There is a trade off with the reflex

design. The speaker system nowresponds like a fourth -order high-passfilter with potentially worse transientresponse.

Response of a reflex enclosure can bedetermined from the followingrelationships.

A=0,./02B=AQts-1 fb(Qifs)C=1 +A+ Valvb)+fb(QifsQts)D=1 /Qts+ fil(Qifs)

For any given frequency f, the relativeresponse in dB (dB=20logR) can befound from,

R -f.4 _cf.,: A).2 (DL2 /3)2

where Is is free air resonant frequency

fn4

Tuning the reflex. The reflex enclosure resonantfrequency is independent of the speaker fittted.Knowing the vent dimensions it can be shownthat the resonant frequency fc is

344.8 Rfc =

2 Vb (L + 1 .7

R= vent radiusL = vent lengthV= cabinet volume

Cc Rc Lc

Equivalent circuitLn,

whereCc =speaker complianceLc =cone mass+air loadFt, =radiation resistance

Cv=boxair compliance4 -vent air massRv=vent radiation

resistance

of the driver, fb is resonant frequency ofthe enclosure and port, Vas is thevolume of air whose compliance isequal to that of the driver, Cas is Q ofthe driver's bass resonance, Q, is ameasure of box losses and can be takenas 7 for normal enclosures, Fn is fifs andvb is enclosure volume.

The resulting curve resembles that ofa fourth -order filter in the case of theoptimum enclosure but closely followsthat of a second -order filter for verysmall enclosures. Small in this caserefers to the Vb/Vas ratio.

A closed box system has astraightforward second -order frequencyresponse in the bass region. This can becalculated from the following.

A=(f/fc)2B=(A-1 )2C=f/(Q0f,)D=C2E=A1(B+D)

fco=A/EN(dB)=201og(1,)

where fc is resonant frequency of thedriver in its enclosure, Q, is theresonance and f is the frequency ofinterest.

When the enclosure is sufficientlysmall to produce a second -orderresponse, the curve is analysed and theequivalent fc and Qo determined. Thisinformation is applied to check theaccuracy of the deduction bycalculating the equivalent infinite -baffleresponse, inverting it and adding it tothe enclosure response. Enclosure sizeis determined mainly by the requiredcut-off frequency.


AUDIO

sources have a bass response that extends to20Hz or lower, at least an octave of musicalinformation is lost.

To make up the shortfall in response,subwoofers are becoming popular.Nevertheless in order to respond down to thelowest octave such systems need to be large.The only sane alternative way of extendingbass response is to use equalisation. I hasten toadd that I am not advocating the use ofgraphic equalisers, but rather a precise inversefrequency response tailored to the systembeing used3.

Probably the best known example of thiskind of system is the sixth -order alignmentproposed by Keele4. The label 'sixth order'refers to the fourth -order roll off of a reflexspeaker whose If output is augmented by asecond -order low pass filter.

Resulting roll off is that of a sixth -orderfilter, as in Fig. 2. Here a reflex cabinet istuned low to produce an over -damped Ifresponse which is then boosted flat by theauxiliary high-pass filter.

Bandpass subwooferWhile experimenting with a similar system Idiscovered an interesting approach to theproblem. I had a Kef B200 mounted in a 19litres sealed cabinet. This 200mm unit has afree -air resonance at 25Hz, a Qt, of 0.51 and aVs of 161 litres. With the speaker mounted inthe cabinet, response of the system is that of asecond order filter with an fc of 90Hz and a Qof 1.8.

The system was equalised for a band-passresponse flat between 30Hz and 100Hz to-3dB. This was made possible by feeding theinput signal through a low-pass second orderfilter, Fig. 3. Inspiration for this system wasthe subwoofer unit presented by Harcourt5.

Although response was well extended,power handling was inadequate. Of course areflex speaker in a small cabinet has a second -order response similar to the closed box. I

reasoned that I could possibly increase powerhandling by fitting a suitable vent andadjusting the equalisation.

The result exceeded my best expectations.Power handling was vastly improved.Moreover the amount of deep bass output wasamazing and without wind noises from thedecidedly small 50mm diameter vent.

Initially I could not account for theperformance of the unit until I read an articleby Plach and Williams from 1951. Ittranspires that when a reflex cabinet is madesufficiently small compared to the driver's V,the phase angle between the vent and driveroutput changes slowly through the ventresonant frequency. The result is that ventradiation is still in phase with the driver'sabove and below the resonant frequency.

This phenomenon accounts for theunequalised response curve which rolls offslowly down to 10Hz and then rapidlyplummets due to the vent and driver radiationbeing in antiphase. Just as importantly thoughthe vent radiation, in a 'normal' reflex designpeaks strongly at the vent's resonant

Lin e i tersT'he Linkwitz-Riley filter is an

I ingenious and versatile networkoriginally designed for use with closedbox speaker systems. Normally, whendesigning such a system the finalresponse curve in the bass region isdetermined by the driver parameters.Unfortunately, with the majority ofdrivers a small enclosure means a highcut-off frequency combined with anundesirable peak in response.

A well designed Linkwitz-Riley filterprovides bass boost and correction forthe peak. The filter is placed before thepower amplifier driving the speakersystem. The net result is a system thatwill respond deeper into the bassregion, with a better transient responsethan that of the original speaker drivenby the power amplifier alone.

In a closed box system the availableacoustic output at low frequencies islimited to the amount of air that can beshifted. This in turn is limited by themaxim:n-1 displacement of the cone.Most modern long -throw drivers arelimited to a linear excursion of 6mmpeak to peak. Typically, damage willoccur at twice this limit so it is vital toensure that this cannot occur. Contrary

to popular belief the amount of powerrequired at low frequencies issubstantially less than the driver'snormal rating. For every halving ofinput frequency the cone excursionwill double for a constant drivevoltage.

Luckily the low frequency content ofmusical signals diminishes rapidlybelow 100Hz. Another objection tobass boosting is that, below resonance,the output is controlled by the linearityof the cone suspension. Howeverresearch has shown that 40% THD isinaudible on program at 40Hz*.

With the design proposed -aseverely overdamped reflex system -efficiency is substantially higherbecause of the contribution of thevent. A feature of reflex systems is thatthe resonant frequency is only slightlyhigher than the free air value sodistortion due to sub -resonantoperation is minimal. What distortionthere is substantially reduced by thehigh impedance air load provided bythe vent.

*This will apply to first harmoniccontent only - Ed.

Linkwitz filter

Design equations

R3

Cnoose f0, 00, fp, Op k= of fp) - (00/0p)/(00/C/p - fp/fo) k < 1

Choose C2: R = 1/6.28 f0 C2 [2 00 (1 + k)] R2=2kR1 C1 = C2 [2 00 (1 + k)12

C3 = C1 (fp/f012 R3 = R1 (f0/fp)2

Max network gain = 40 log (f0/fp)

k>1 required

-P

fo, Oo

A = Linkwitz filter responseB = Speaker responseC = System response


AUDIO

dB

+20 -

15

+10 -

5

0

5

-10

-15

-20

-2510 100

Hz

1000

Fig. 3. Reducing the size of the reflex enclosureproduced second -order characteristic B. Addingelectronic second -order low-pass transferfunction A produced a bandpass systemexhibiting curve C.

frequency. With the system described here theQ of this peak is much reduced, Fig. 4. Itfollows that the vent can be made smallerbecause of the lower sound velocity atresonance without generating wind noise. Ihave found that a vent area 1/16 of the driverradiating area suffices.

Just as important is the inherently low rateof roll off in the system's stop band. We havea reflex system with the roll off characteristicsof a sealed box and with it the better transientresponse of a sealed enclosure system.

Building a full range system based on theprinciple proved to be quite simple. All that isneeded is a Linkwitz-Riley filter3. Thisnetwork is designed to compensate for a givensecond -order response and substitute it withanother of the designer's choosing.

Originally the Linkwitz-Riley network wasdesigned for use with closed box systems.However its second -order response curve iseasily calculated. All that remains beforedesigning the network is to determine theequivalent fo and Q of the unequalised systemresponse.

The system has the most desirablecharacteristic of a sealed speaker cabinet -superior transient response. Since the vent andenclosure need to be small to obtain thecorrect response, the cabinet can be madeunobtrusive. It has an extended low frequencyresponse combined with low cone excursionsdue to the favourable phase shift betweendriver and vent. Because the final responsecurve is equalised by an external filter, driverswith relatively highfs and Os - i.e. cheap ones- will suffice.

Full -range system detailsIn my design, called the Microflex, I used apair of Kef 8200s for bass and mid rangetogether with a pair of Audax DTW 100 T125F FFG tweeters.

My enclosure is made from 15mm mediumdensity chipboard panels with overalldimensions of 355mm high by 258mm wideby 303mm. This yields an internal volume of

19.961. The cabinet is unlagged and tuned to34Hz by a 2I5mm long 51mm diameter vent.

The Kef speakers have a high frequency rolloff above 3.5kHz accompanied by a ratherlarge response peak. Roll off is that of asecond order filter with a Q of 3.5. I haddecided from the outset that the speakersystem was to be active so I turned this roll offinto an advantage. I fed the bass signalthrough a 3.5kHz low pass second order filterwith a Q of 0.143. This combination gives anacoustic response from the B200 of a fourthorder low pass with a Q of 0.5.

This is ideal since the aim was to implementa Linkwitz-Riley-Riley crossover. This type ofcrossover uses fourth order filtering for bothhigh and low pass sections combined with a Qof 0.5 for the best possible transient response.My tweeters are titanium dome types with anatural resonance frequency of 1.2kHz whichis over an octave below the crossoverfrequency. Filtering for the tweeter is providedby a cascaded pair of Sallen and Key second -order high pass filters built around a dual op -amp, Fig. 5.

At the bass end, the input signal is firstbuffered before being processed. Active filtersonly operate as intended when driven from alow impedance source. The signal is then fedinto a second -order high-pass filter with aturnover frequency of 30Hz and a Q of 1. Thisfilter is an optional extra and is most usefulwhen using vinyl discs.

Bass information is now fed to the Linkwitz-Riley filter. Values shown are determinedempirically. From here the signal is fed intothe low pass filter previously described, whichforms part of the low pass crossover network.Also fed from the input buffer are thecascaded high-pass crossover filter sections.From here, output is fed to the poweramplifier via a preset potentiometer to equalisesignal levels to the speakers.

Bass response is flat down to 30Hz and thesystem can produce a good acoustic sine waveat this frequency. Transient response is goodand system performance is good on allmaterial - especially speech. I havedismantled my subwoofer as it is no longernecessary. I now feel that the real historicalsignificance of Thiele/Small theory will beshowing us where to set our equalisers.

ReferencesI. Thiele, A. N., Loudspeakers in vented boxes,J. Aud. Eng. Soc.,Vol 19. May/Jun 1971.2. Small, R.H., Closed box loudspeakersystems, J. Aud. Eng. Soc. Vol 21, Jun/Sep19733. Linkwitz-Riley, S., A three enclosureloudspeaker system, parts 1-3, LoudspeakerBuilder 2-4, 1980.4. Keele, D. B., A new set of sixth -order ventedbox loudspeaker system alignments, J. Aud.Eng. Soc., vol 23, pp354-360 June 1975.5. Harcourt, R. I., An acoustically smallloudspeaker, Wireless World, Oct. 1980.

10

+20

0dB

-20

dB

30°

60°

90°

120°

150

100Hz

Hz -

1000

Vent o/p reflex

Vent o/p Microflex

Hz

Vent phaseMicroflex

Vent phase reflex

04 1.0

f/Fb

(a)

(h)

tt I

16

Fig. 4. Curves for the electronically augmentedreflex enclosure show that response is flat downto the cut-off point.


AUDIO

Bass equalisation

ValuesResistors, 1%R1,6,7,10,135 6 kR2,4 12kR3,5 22kR8,9 15k

811,12 100kR14,15 820kR16,17 4k7

R2

CapacitorsC1-4 2n7C5,6

C7,8C9

C10C11

012,13C14

100n4n710n3n333n100n cer68n

R4

9mC

To tweeterpower amp

R13 C8

R14

R15

SemconductorsA1-6 TL072 (3off)IC2 7815IC3 7915

C14

+15

C12 7OV

C13

15

IC2

C10

IIC3 I

R16 R17\AA,A,--\

011A4

To wooferpower amp

Fig. 5. Linkwitz-Riley crossover withelectronic woofer equalisation. This circuitcompensates for low -frequency responsedeviation in a small reflex enclosure -Microtlex. Further optional filteringremoves sub-sonics for disc replay.

LOW COST RANGER1 PCBDESIGN FROM SEETRAX

Circuit SchematicCircuit CapturePCB DesignHost Of Outputs

All -In -One Design System£100

Fully Integrated Auto Router£50

Ask Us About Trade -In DealsCall Now For Demo Disk on 0705 591037

Seetrax CAE Hinton Daubnay HouseBroadway Lane Lovedean Hants P08 OSG

Tel: 0705 591037 Fax: 0705 599036

REDUCED

PRICE!

Pr

WhatThe Press

SaidAbout1241VGzRiFor

ost small+isers.

SeetraxRanger]

providesasystem

at an,

affordableprice.

It is betterthan

EasyPCor Tsien'sBoardrnaker

sincetprov;desa /ot more

andtakes

the designall the

Way

from schematictoPCB

- otherPacks

separatedesigns

for both,that is, no schematiccapture./t is more expensive

but the abilityto draw

in thecircuit

diagramand quick/V

turnit into

a boarddesign

easi/ymakes

up for this.SourceJUNE /99/acticalElectronics

Pay by Visa or AccessWOE

101=1M


February 1994 ELECTRONICS WORLD + WIRELESS WORLD 1 1 1

AUDIO

The spirit ofBASS

HzIrd and fast lines cannot be drawn, butbass energy in music starts somewherearound 300Hz, contains over half the

octaves that music spans, and turns into sub -bass - something you feel more than hear -below about 30Hz.

Bass energies have been important tomankind for more than 10,000 years. The rockfestival that can be heard miles away useskilowatts of electricity to parallel tribal drum-ming ceremonies, audible at over ten miles.But the lowermost three octaves, 10-80Hz, arerarely appreciated in the majority of Westernlistening, which takes place on compromised,so called high-fidelity systems.

While it is superficially possible to appreci-ate much 'classical' Western music largelywithout these lower registers, the fulsomereproduction of bass profoundly enhances andexpands the experience and consciousness ofthe serious music lover. With an 18th centuryoratorio or a pipe organ, the low content mayonly be the grandiose. But most of rhythmicmusic's therapeutic power resides in the lowbass.

There is a widespread misconception amongthose without a global appreciation of musicthat very low bass is somehow dangerous*. Ihave repeatedly experienced outdoors some ofthe highest musical low -bass sound pressurelevels yet created. For myself and many oth-ers, the experience is extremely physical, plea-surable and ultimately cathartic.

Bass - the acoustic backgroundRudimentary audio knowledge talks of musicrequiring a -3dB bandwidth that begins at20Hz or even higher. It is true that few tradi-tional instruments generate fundamentalsbelow 40Hz, but this does not mean that they

* 1 believe the misunderstanding arises from LyallWatson's first booku in which he reported howexperiments with continuous subsonic waveforms -not musical 'sub -bass' - demonstrated destructiveand fatal qualities.

Unlike traditional Westernmusic producers, manymodern artists rely heavilydeep bass for exciting theiraudiences. Ben Duncandiscusses why U2 neverquite sounds the same inyour living room.

produce no output. Moreover, pipe organs,gongs and synthesizers, as well as diversesounds that have been sampled and pitchshifted, as used in today's House and Ambientmusic, reach down to at least 20Hz, if notbelow 10Hz.

In domestic reproduction, the relevance offrequencies below 100Hz and the ability torecreate them in any ordinary living room isdebatable. In many cases, they are not realis-able on at least three counts. First, the LFresponse of two-way domestic speaker sys-tems is stymied by acceptable size. Second,the maximum sound pressure level that can bedeveloped in the bottom decade of domesticspeakers is usually severely limited by driverexcursion limits. Particularly in sealed boxenclosure, the cone moves proportionately fur-ther with decreasing frequency for a givensound pressure level.

A continuum of trade-offs exists betweenpiston arca, excursion and allowable harmon-ic, Fig.1, and Doppler distortion. So the aver-age domestic speaker capable of 105dB at1kHz is likely to be limited to 70dB or lessbelow 50Hz. A speaker that has been tuned togo lower will have less maximum output sinceextension is traded against excursion.

Third, and often overlooked, Robinson andDadson curves, Fig. 2, show that the thresholdof human perception at 20Hz - the minimumaudible frequency - is around 75dB pressurelevel. This suggests that a speaker has to pro-duce more than 70dB at 20Hz for programmeat this frequency to even begin to be audible.

On the other hand, a number of specialistbass enclosures and sub -woofers are capableof producing useable bass to well below 40Hzwith dimensions suitable a domestic settingsand smaller than elementary theory suggests ispossible.

It is sometimes thought that our perceptionof bass qualities in familiar music is greatlyinfluenced by the ear's ability to synthesisemissing or attenuated frequencies using the


AUDIO

mid frequency harmonics as cues. It followsthat accurate mid -frequency reproduction hasas much influence on bass sonics as the bassper se.

Synthesis of the fundamental explains howmini and micro -monitors and the majority ofsmall domestic speakers can satisfactorily por-tray bass lines while being incapable of repro-ducing the fundamental audibly. Still, themind has to work to achieve this, and some ofmusic's higher qualities are lost.

Physical presence of the lower octaves, say-3dB at or below 50Hz, is found more relax-ing and preferable to most listeners. Equally,deficiencies, excesses or imbalances in the HFare inversely reciprocated by our senses:Listeners may hear an excess of low bass as alack of treble, and rice-versa2.

It is also not widely appreciated that the ruleof thumb that 1dB is the smallest perceptiblechange to the average listener is true only atconversational pressure levels, at mid -fre-quencies. Smaller changes are exaggerated athigher and particulary lower frequencies. Atmid -frequencies, a 10dB increase from 65dBto 75dB is perceived as a doubling of loud-ness, but the same change at 20Hz would beperceived as more than a loudness quadru-pling.

This is evident from Fig. 2, by following thedistance between the curves. It follows thatwhereas electronics and audio both use log 20decibel scales for discussions and comparisonsof attenuation, gain and amplitude, the mean-ing of audio equipment figures in decibelsmust be considered.Ambient sounds are important cues, and whilenot musical per se, they are part of the holo-gram of up-market hi-fi reproduction. Theyinclude the 'feel' - almost a throb - of thewooden stage or floor that the musicians areon. These are difficult to describe, but obviouswhen present.

System roll -offs, capacitors and delaysMost audio equipment contains too many dcblocking capacitors. The only positivelymandatory series capacitors in audio are onefor analogue disc, usually RIAA/IEC, and onefor each standard of tape equalisation. Atmost, up to three series capacitors are essen-tial. These are for up to -18dB/octave high-pass filtration to roll -off unwanted bass andeven these should be preferably bypassable.

Every series capacitor in the chain creates ahigh-pass pole. As a result, the response fallsoff below xHz at -6dB/octave times the num-ber of series capacitors - assuming all roll -offsare the same within an octave or two. Sincefiltering caused by dc blocking is wholly pas-sive, it is at least highly damped (sub-Bessel).

With active high-pass filters, non linearitiesin degrees phase per hertz (with frequencyplotted linearly) are compounded by Q andtheir gross effect reaches higher into the audiospectrum with every pole.

Setting very low -3dB roll -off points, wellbelow 20Hz, and minimising the number ofroll -off as is important on several counts. Oneis group delay or phase dispersion. In 1990, I

published possibly the first analysis of theeffects of audio signal chains in the time andfrequency domains3.

Figure 3 shows a portion of If music pro-gram variably delayed by 0 to 8ms. The delaydepends on the dominant component after thesignal has passed through each of 24 stageshaving a modest 3dB low-pass roll -off, or fm,

40

30

20

10

0

of 2.1Hz. Such response is typical of a con-ventional, complex recording studio signalpath, so it is a feature of most recordings. Itwas also a feature of earlier hi-fi systems.Fortunately not everyone is sensitive to it.

A well designed modern domestic replaysystem can have as little as one series capaci-tor, giving an ultimate roll -off in the electron-

-

- Single driver

1\i\

Double driver

Quad driver

:4........._z..,....,_,...-7,-:,..T.,..........z.p,

10 100

Logarithmic Frequency in Hz

Fig. 1. Cone excursion required to produce 135d8 sound -pressure level. Even using four 15in driversthe excursion needed at 40Hz is considerable.

120

100

80

60

40

20

0

1000

-77.7 iLOUDNESS

120 LEVEL (PHON)

OpPer.

di.POak

ito---T

too

.

18,

V

90

R11110/11M I

1

80, _

70 Anliikk -"IWAlfrillik60\

50\X 7 - 1r

1

\100

Ni

30 ii11811/1,1A

I^.47iNikki._ 120 -."11111M40 V

7.111111111111111Villgri..'

...". '-.

10

-. '"- --..

OM111/.41MOIV /

ri VjMAF -7.

CURVE''' I

___. - I II20 100

FREQUENCY IN HERTZ1000 5000 10 000

Fig. 2. Often overlooked equal loudness contours from Robinson and Dadson show that a speakerhas to produce over 70dB to be detectable at 20Hz.

February 1994 ELECTRONICS WORLD + WIRELESS WORLD 11:3

AUDIO

INPUT

(Volts)

1.00

0.80

0.60

0.4

0.20

0.00

-0.20

-0.40

-0.60

-0.80

-1.000 60

STM=2m, MXc=594, RELer=2e-3

HPLPCH-6 Temperature= 27 C.BOR 04.12.90

T in M

Complex LF music signal at output of studio chain model:

OUTPUT

120 180

I in M illiseconds

ics limited to -6dB/octave. Phase dispersionembedded in the recording cannot be fixedthis way, but further damage is avoided. Inrecording studio equipment chains, and wher-ever more than one series capacitor is essen-tial, phase dispersion is kept at bay by settingF3L« I Hz.

Properties of real capacitors lead to the sameconclusion. An electrolytic capacitor can actas a filter with a huge temperature coefficient,frequency -dependent capacitance and com-plex, non-linear parasitics4-5. The thresholdwhere they cease to act this way is properlyset by considering psychoaccoustics, takingthe outer limits, and adding a margin of 300%(just 10dB, remember).

For dc blocking in purist audio equipment,electrolytic capacitor values are practically as

Music - the ultimate test signalMusical information spans at least ten octaves,while mankind's entire visual world isdrawn from just the one octave of visible light.

Music signals are the antithesis of the puresine wave routinely used to test audioequipment. The sine wave has a crest factorof 3dB. The most adventurous test signalused for the majority of 'objective' audiotesting is filtered and slightly clipped pinknoise. This can crudely imitate music'svariable crest factor or 'peak -to -mean ratio',which averages 15dB but varies fromaround 8dB up to 25dB.Music is made up of 'shifty' sine waves,many of them harmonically related. Theirperiodicity, amplitude, frequency and rateof change of these parameters, henceenvelope, is some combination of theregular, the rhythmic, the self -similar, thenested, the random and the chaotic.

240 300

ment. To astute listeners equipped with a suit -OUTPUT able system, a plausible 1dB change in bass(Volts) response as the system warms up will be per -

1.00ceivable as around a doubling in loudness. Butif net F3L is 16mHz, then the same shift will

0.80 lie below audibility.

0.60

0.40

0.20

0.00

-0.20

-0.40

-0.60

-0.80

-1.00

MT=3e-5, MVe=2m.

C

Fig. 3. Phase dispersion illustrated as varyingdelay in the output of a studio signal chain,compared to the input. Delay varies up to 10mswith dominant instantaneous frequency. In afully direct coupled system, the delay would beabsent.

large as can be borne by the circuit -board areaavailable, while maintaining a resonant fre-quency that is comfortably above 20kHz. Inthe few places I use DC blocking, an audiograde, low inductance 10000 electrolyticcapacitor is combined with 101(0, giving anF3L of 0.016Hz.

Why not use film capacitors, whose charac-teristics are nearly perfect ? Well, to attaineven a modest F3L of say 1Hz using moderateresistor values, the values needed are unac-ceptably physically large and expensive -16µF, for 10ka A smaller and more accept-ably priced 150nF would require partneringwith IMO for an F3L of 1Hz. But a resistancethis high will cause problems with micro -phony, electrostatic pickup, dc offset and pos-sibly thermal noise.

Microphony is by far the most serious prob-lem. The capacitor allows a distorted acousticsignal to egress at a random point in the chain,causing tonal defects and smearing. This is acommon defect in vacuum tube amplifiers,with their high impedance networks. In dis-crete transistor topologies, some capacitorlocations such as bootstraps can cause biasshifts and 'tails' when driven or momentarilyoverdriven. Asymmetrical music signals cancause this even though the system may behaveperfectly with a sinewave (see panel).It is rarely acknowledged that the temperaturecoefficient of resistors and particulary thecapacitors that set F3Ls will cause the systemF3L, as well as the phase dispersion, to changeas a large audio system warms up. Assumethat the net system F3L is set too high at 30Hz- a common feature of mass -market equip -

Direct couplingIn the past decade, the construction of direct-ly coupled audio circuits has been made pos-sible by the arrival of op -amps that combinegood ac performance with high DC precision.PMI's OP37 was the first, followed by LinearTechnology's LTI 037 and LTI 028 in 1983.

II this was made possible by George Erdi'swork in combining bias -current cancellationwith low -noise input pairs6a. Since then,Analog Devices/PMI, Harris, and Texas haveall produced even better parts. But what aboutdrift ? My AMP -01 preamplifier design of19838 contained up to five direct -coupled op -amps in a line. These were individually nulled.After ten years, I can report that drift causingdc levels to build up, to make pots `scrapey'or cause significant clicks when switching, hasnot been an issue.

Since then, I have designed a crossover foruse in some of the world's foremost recordingstudios. It has twelve direct coupled stages andrequires absolute reliability over thousands ofhours operation. Using Harris HA3 -522I-5s toreplace the NE5534s, no nulling was required.In addition, bass sonics are vastly improvedover the MkI design, which was mostly accoupled. Servos may be used to force offsetsto zero. But this technique just moves thecapacitor, adds another, and adds an op -ampwith its attendant cost, supply consumptionand noise injection.

Power Supply optimisationBenefits of regulated supplies with regard tolow frequency capability and purity have beencovered in some depth9". Considering enve-lope modulation, the 'sample rate' of a con-ventional 50/60Hz power supply interacts withthe bass frequencies dominating the musicenvelope.

In 1986, I helped create the DVT-300. Thiswas the first high power - 600W per channel- audio amplifier to use a quasi -resonantswitching supply. Switching was at 80kHz,which placed the power supply recharge ratewell above the audio band.

Today, several UK amplifier makers use thisand other power conversion techniques above20kHz. Bass clarity has been shown to beimproved by exchanging a given output stagecircuit between linear and 80kHz switchingpower supplies.

Dynamic tonal correctionKnowing about the ear's non-linear perceptionof the 3D change-of-SPL versus SPL versusfrequency, instead of frequency leads to theconclusion that sonic accuracy in audio sys-tems hinges on reproducing music at precise-ly the original sound pressure level (this sameeffect causes musicians think in terms ofpitch). At any other level, the music will be


AUDIO

Bypass loop

Hot 0

Cold 0

Balanced audioinput

RFfilter

Bal. to unbal.

3

Audio linesensing

derivationWeightedsummer

Powersupply

EQ network

Substratedriver

Mute relay

Line driver Bypass relay

T ue RMS converter

Differentialreference generater

± 16 V to allAnaloguecircuitry

tonally and dynamically different, and the lowbass content will be most affected by the mostcommon practice of listening to reproducedmusic at 10 or 20dB below the original.

On this basis, turned -down volume controlsact to make the music recede into the distance,rather than simply making the musiciansshrink. Small tonal differences can masquer-ade as dynamic differences. The 'loudness'switch on down-market audio equipmentattempts to compensate for this. It increasesLF (and HF) at low levels, but its operation isbased for the most part on false premises".

At the very least, any tonal compensationfor the purposes of replay at lower than reallevels requires some reference to the originalsound level. The correct equalisation is cer-tainly not the Fletcher and Munson curve. In1986, I designed the Infle.vor - an automatictonal compensator, Fig. 4, which was morecorrectly referenced psychoaccoustically12.

While the curves accepted professionally"most serious domestic listeners, preferring theminimum signal path, either compensate forthe tonal errors in their heads, or listen at whatseems to be the original level.

DynamicsWith the exception of COG ceramic capaci-tors, all components used in analogue elec-tronics vary in characteristics due to the tem-perature changes that occur in real equipment.

None is more acutely sensitive and fastresponding than the semiconductor, particu-lary one with small junctions. Conventionally,thermal distortion is considered to be anaffliction restricted to power amplifiers at bassfrequencies. By bass I mean signals that canhave periods long enough, above say 10ms, tocause considerable cycle -to -cycle and evensub -cycle temperature changes.

Higher level errors generated by inter -com-ponent and inter -stage thermal feedback aremostly limited to monolithic ICs. But even

Ch. 1 Ch. 2 Ch 3

Binarybuss

0

0

BAC controlmatrix

Quadrantconverter

Differentialsummer-subtractor

DBC converter -

Fig. 4. The Inflexor, a dynamic loudnesscompensator, allows music to retain its bassvitality when reproduced at up to 40d8 belowthe original sound -pressure level. The shortaudio signal path can be seen at the top, andbelow, the analogue computing elements usedto track compensatory equalisation. Thisovercomes the ear's non-linear perception ofchange-of-SPL versus SPL versus frequency.

small signal transistors in long tailed pairsreflect their thermal environment n the sonicend -result".

Thermal distortion can be measured withsine waves on an open -loop fixture°, but itsreal effect on music is more complex, as it canbe highly asymmetric (see panel). So forexample, one half of an output stage willexperience greater cycle -to -cycle changes,creating an error that manifests as even -har-monic residue. Large geometry transistorswith thermally conductive epoxy casing havebeen used to reduce the audibility of thermaltails.Both analogue and digital ICs can be as sen-sitive to vibration, by bass in particular, ascapacitors. Spectral analysis has demonstratedthe value of shock mounting a CD player'sdigital electronics. In a UK university, laserinterferometry has showed that the legs ofsemiconductor devices dance when passingmusic, but whether this is caused by acousticexcitation, or by magneto -restriction effects, isas yet unclear. Bob Pease's anecdote° aboutintegrated circuits being transparent to infraredradiation and hence modulation by 50Hzambient lighting flicker or noisy leds, is anexample of the potential for left -of -field basscontamination.

0 Audiooutput

Overrange>125 dBC

LEDbar

readoutcorresponds

to dB SPL

<75 dBC

LED drive array

DBC: Digital to binary convertorBAC: Binary to analog convertor

References1. Butler, T., In search of 3D sound, Hi-FiNews, Nov '93.2. Philip Newell, Hi-fi & monitoring: hearing isbelieving, Studio Sound, Oct 1993.3. Duncan, B., The signal chain, Studio Sound,June 1991t.4. Duncan, B., With a Strange Device, Hi-FiNews, Parts 2-5, Jun, July, Sept, Nov 1986.(Available in one volume, as Audio QualityInvestigations, Vol.1 from Audio Synthesis, 99Lapwing Lane, Manchester M20 OUT. Tel.061 434 0126.)5. Hawksford, M., The Essex Echo, tracks 3 &4, Hi-Fi News, Oct 1986 & Feb 1987.6. Erdi, G. & Cakhnokhi, Y., A bipolar op -ampwith a noise resistance of less than 50 ohms,IEEE solid-state circuits conference, Feb 1986.7. Erdi, G., Jung, W., Bernardi, S., Schwartz,T., Op -amps tackles noise, Electronics DesignInternational, Dec 1980.8. Duncan, B., AMP -01 modular preamplifier,Hi-Fi News, May -Nov 1984.9. Duncan, B., PSU regulation boosts audioperformance, EW+WW, Oct 1992.10. Hawksford, M., The Plot Thickens,EW+WW, Letters, Mar 1991.11. Holman, T., & Kampmann, F., LoudnessCompensation: Use & abuse, Journal of theAES July/Aug 1978.12. Duncan, B., Dynamic LoudnessCompensation, Reproduced Sound 3, Nov1987, Institute of Acoustics.13. Dibble, K., Greystone Inflexor - psycholo-gy of volume restriction, Disco & Club TradeInternational, Feb '88.15. Graeme, J. G., Designing with op -amps,application alternatives, pp. 229-231, McGrawHill, 1977.16: Pease, R., Troubleshooting analogue cir-cuits, Butterworth, 1991.17. Watson, L., Supernature -a natural historyof the Supernatural, Hodder & Stoughton,1973


2816A-30 HOUSE MARKED2Mb SIMMS FOR IBM RS6000 RISC MACHINEIBM PART NO. 68X6271 PANASONIC PART NO.MN4B40512S85 512K X 40TMS9000NL PROCESSOR £20TMS9901/2/3NL, TIM9904NL £4.50 eaHM6167LP-8 65pKM28C65-25 SAMSUNG 89/90 1100 AVAILABLE £3

REGULATORSLM338K £6

£2LM323K 5VA METAL £3SANKEN STR451 USED IN AMSTRAD MONITORS £578H12ASC 12V 5A £578M05 5V 0.5A 7/£1LM317H T05 CAN £1

ElLM317 METAL £2.207812 METAL 12V lA £17805/12/15/24V plastic 25p 100+ 20p 1000+ 15p7905/12/15/24 plastic 25p 100+ 20p 1000+ 15pCA3085 TO9 2/E1

ITH RESET OUTPUT Elea C-50/100

CRYSTAL OSCILLATORS2M4576 3M6864 5M0 5M76 6M144 7M000 7M37 8 8M00012M000 14M3181 17M6256 16M257 18M000 20M000 23M58724M000 25M175 27M0 27M036 28M322 32M000 35M481640M000 44M4444 44M900 48M000 64M000 1M000 1M84324M000 10M000 16M000 18M-452 0038M10000 56M6092 76M1 84M0 ch

TALC ---256 10M368 17M6256 18M432 25M000 28M4694 31M4695 --

48M000 55M500 111M80 112M80 114M318 114M80 IMO ;3M58564 3M93216 4M000 4M19304 4M433619 4M608 4M91s59.1M8432 2M000 2M4576 2M77 3M00 3M2768 3M579545

5M000 5M0688 6M0000 6M400 8M000 8M488 9M8304 10M24010M245 10M70000 11M000 12M000 13M000 13M270 14M00014M381818 15M000 16M000 16M5888 17M000 20M00021M300 21M855 22M1184 24M000 34M368 36M7562536M76875 36M78125 36M79375 36M80625 36M8187536M83125 36M84375 38M900 49M504 54M19166 54M741657M75833 60M000 69M545 69M550 BN 26M995 RD27M045OR27M095 YW27M145 GN27M195 BL27M245 £1

COMPUTER ICSP8271 BBC DISC CONTROLLER CHIP EX EQPTSAA5050 TELETEXT CHIP EX EQPT £52817A-20 (2K x 8) EEPROM ex eqpt.27C64-25 used/wiped £1.50 100+ £1D41256C-15 2561(xl PULLS 9 FOR £580C31 MICRO £2P8749H MICRO £508751-8 NEW £10MK48Z02-20 ZERO POWER RAM EQUIV 6116LP £4NEW 4164-15 £1USED 41256-15 ClUSED 4164-15 60pBBC VIDEO ULA £108051 MICRO £1.25KS82C55-250 SAMSUNG 89/90 1100 AVAILABLE £39 x 41256-15 SIMM £108 x 4164 SIP MODULE NEW27128A 250ns EPROM USED £2 NEW £2.3027C1001 -20Z NEW IM EPROM £6FLOPPY DISC CONTROLLER CHIPS 1771 £16FLOPPY DISC CONTROLLER CHIPS 1772 £17.5068000.8 PROCESSOR NEW £6HD6384-8 £5ALL USED EPROMS ERASED AND BLANK CHECKED2716-45 USED £2 100/£12732-45 USED £2 100/£12764-30 USED £2 100/£1.6027C256-30 USED £227C512 USED1702 EPROM EX EQPT £52114 EX EQPT 50p 4116 EX EQPT 70p6264-15 8k STATIC RAM £2GR281 NON VOLATILE RAM EQUIV 6116 £5Z80A S10-0 £1.25TMS27PC128-25 ONE SHOT 27C128 El ea 100/E7080387-16 CO PROCESSOR (OK WITH 25MHz 386) £357126 31/2 DIGIT LCD DRIVER CHIP £2 ea

E2£65 ea

LM323K 5V 3A PLASTIC

LM3177 PLASTIC T0220 variable

TRANSIST RSMPSA922N2907ABC477, BC488BC107 BCY70 PREFORMED LEADSfull spec £1 E4/100 £30/1BC557, BC238C, BC308B El /30 £3.50/1002N3819 FETS short leads 4/£1

POWER TRANSISTORS0C29 £1.35 ea

3/£12/£1

3/E1 100/E222/E12/£1

100/£3510/E14

P POWER FET IRF9531 8A60VN POWER FET IRF531 8A60V2SC1520 sim BF259TIP 141/2 El ea TIP 112/42BSE9301 100V 1DA DARL SIM TIP121PLASTIC 3055 OR 2955 equiv 50p2N3773 NPN 25A 160V £1.60

TEXTOOL ZIF SOCKETS28 PIN USED £3 40 PIN NEW £10SINGLE IN LINE 32 WAY CAN BE GANGED FOR USE WITHANY DUAL IN LINE DEVICES COUPLING SUPPLIED 2/£1.50

QUARTZ HALOGEN LAMPS12V 50watt LAMP TYPE M312 5.1 ea HOLDERS 60p ea24V 150 WATTS LAMP TYPE AI £2.50 each

MISCELLANEOUSMINIATURE FERRITE MAGNETS 4x4x3mm 10/£1

KEYTRONICSTEL. 0279-505543FAX. 0279-757656

P 0 BOX 634BISHOPS STORTFORD

HERTFORDSHIRE CM23 2RX

100nF 63V X /R PHILIPS SURFACE MOUNT 30K available£42/4000 box

10NF 63V X7R PHILIPS SURFACE MOUNT 160Kavailable £30/4000 boxETHERNET 4 PAIR TRANSCEIVER CABLE. BELDEN TYPE9892 £60 for 50 metresSWITCHED MODE PSU 40 WATT UNCASED QTY.AVAILABLE +5v 5A, +12V 2A, 12V 500mA FLOATING

£9.95 (E2)SIM RS 361-018 Each pair foil screened + overall braidedscreen £300 for 305 metre drum+ plus PVC outer sheath. ON 305 metre drums330nF 10% 250V X2 AC RATED PHILIPS 60K

220R 2.5W WIREWOUND RESISTOR 60K AVAILABLE £20/100AVAILABLE

CMOS 555 TIMERS00/1000

2/£12/3 LITHIUM CELLS AS USED IN COMPACT CAMERAS £2.50ea.ICM7126CPL CMOS 31/2 DIGIT LCD DRIVER CHIP E2ea36 CORE 7/0.2mm OVERALL SCREENED £50/100mLITHIUM CELL 1/2 AA SIZE 2 FOR £1PASSIVE INFRA RED SENSOR CHIP + MIRROR + CIRCUIT

EUROCARD 21 -SLOT BACK PLANE 96/96 -WAY£2 each

£25 ea"PROTONIC 24 VARIBUS" 16.7"x5" FIBREGLASSMULTILAYER PRH PCBEUROCARD 96 -WAY EXTENDER BOARD 290x 100mm

£.10 ea"PROTONIC 24" c/w 2 SUPPORT ARMS/EJECTORS.DIN 41612 96 -WAY A/B/C SOCKET PCB RIGHT ANGLE £1.30DIN 41612 96 -WAY A/I3/C SOCKET WIRE WRAP PINS £1.30DIN 41612 64 -WAY NC SOCKET WIRE WRAP PINS ...... £1DIN 41612 64 -WAY NC PLUG PCB RIGHT ANGLE £1DIN 41612 64 -WAY A/B SOCKET WIRE WRAP (2 -ROW BODY)

£1BT PLUG +LEAD 3/£113A MOULDED PLUG+2m lead £1MIN. TOGGLE SWITCH 1 POLE do PCB type 5/E1LCD MODULE sim. LM018 but needs 150 to 250V AC for display40 x 2 characters 182 x 35.13mmTL431 2.5 to 36V T092 ADJ. SHUNT REG 2/£16-32 UNC 5/16 POZI PAN SCREWS E1/100

NUTS £125/100PUSH SWITCH CHANGEOVER 2/£1RS232 SERIAL CABLE D25 WAY MALE CONNECTORS

£5.90 ea (£1.30)25 FEET LONG, 15 PINS WIRED BRAID + FOIL SCREENS

INMAC LIST PRICE £30LCD DISPLAY sim Hitachi LM016L £6.50AMERICAN 2/3 PIN CHASSIS SOCKET 2/£1HUMIDITY SWITCH ADJUSTABLE £2WIRE ENDED FUSES 0.25A 30/£1NEW ULTRASONIC TRANSDUCERS 32kHz £2/prAlso available28 slot van -bus backplane same size +Price NEWPOWERFUL SMALL CYLINDRICAL MAGNETS 3/£1BNC 500HM SCREENED CHASSIS SOCKET 2/£1SMALL MICROWAVE DIODES AE1 0C1026A 2/£1D.I.L. SWITCHES 10 -WAY Cl 8 -WAY 80p 4/5/6 -WAY 80p180VOLT 1WATT ZENERS also 12V & 75V 20/£1VN1OLM 3/£1MIN GLASS NEONS 10/E1RELAY 5V 2 -pole changeover looks like RS 355-741 markedSTC 47WBost £1 eaMINIATURE CO -AX FREE PLUG RS 456-071 2/E1MINIATURE CO -AX FREE SKT RS 456-273 2/£1.50DIL REED RELAY 2 POLE No CONTACTS ftPCB WITH 2N2646 UNIJUNCTION WITH 12V 4 -POLE RELAY

El00m 0.5W thick film resistors (yes tour hundred megohms) 4/1

TRAIN GAUGES 40 ohm Foil type polyester backed balco gridalloy

CTRET MICROPHONE INSERT£1.50 ea 10+ £1

£0.90Meer Hall effect IC Micro Switch no 613 SS4 sim RS 304-267

£2.50 100+ £1.50HA. EFFEOTICTIGS0 -7- £1OSC LOSCOPE PROBE SWITCHED- x-1 .x1-0 -.£1.21 pole -way rotary switch 4/£1AUDIO S LM380 LM386 TDA 2003 El ea555 TIM RS £1 741 OP AMP 6/£1ZN414 RADIO CHIP 80pCOAX UGS nice ones 4/£1CO =ACK TO BACK JOINERS4x- EMBRANE KEYBOARD

UCTOR 20pH 1.5A1.25" PANEL FUSEHOLDERSCHROMED STEEL xF41-4-4- 1" orc12V 1 w e lamps fit most modern cars

EO CASSETTE HEADMONO CASS. HEAD £1 ERASE HEADTHERMAL CUT OUTS 50 77 85 120°CTHERMAL FUSES 220°C/121°C 240V 15ATRANSISTOR MOUNTING PADS TO -5/70-18

(TOR COVERS

Et. 100n 600V S .13/£]i --'10n 250V radial 1 Omm

1N540/ 3A 1000V 8/£11N4148 100/0.501N4004 SD4 1A 300V 100/E31N5401 3A 100V 10/£1BA158 lA 400V fast recovery 100/0BY127 1200V 1.2A 10/£1BY254 800V 3A 8/E1BY255 1300V 3A 6/£16A 100V SIMILAR MR751 4/£11A 600V BRIDGE RECTIFIER 4/E14A 100V BRIDGE 3/£16A 100V BRIDGE 2/E110A 200V BRIDGE £1.5025A 200 V BRIDGE £2 10/E1825A 400V BRIDGE £2.50 10/£222KBP02 IN LINE 2A 200V BRIDGE REC 8/E1

SCRSPULSE TRANSFORMERS 1 1+1 £1.252P4M EQUIV C106D 3/£1TICV106D 800mA 400C SCR 3/E1 100/E15MEU21 PROG. UNIJUNCTION 3/£1

TRIACS DIACS 4/E1NEC TRIAC ACO8F 8A 600V-1-0220 .... 5/£2100/£30TXAL225 8A 500V 5mA GATE 2/E1 100/E35BTA 08-400 ISO TAB 400V 5mA GATE 90pTRAL2230D 30A 400V ISOLATED STUD £5 eaTRIAC 1A 800V TLC3817 16k AVAILABLE

CONNECTORS

5 FOR £1 £15/100

D25 IDC PLUG OR SOCKET34 -way card edge IDCCONNECTOR (disk drive type)CENTRONICS 36 WAY IDC PLUGCENTRONICS 36 WAY IDC SKTBBC TO CENTRONICS PRINTER LEAD 1.5MCENTRONICS 36 WAY PLUG SOLDER TYPEUSED CENTRONICS 36W PLUG +SKT14 WAY IDC BLOCK HEADER SKT.

PHOTO DEVICESHI BRIGHTNESS LEDS CQX24 RED 5/E1SLOTTED OPTO-SWITCH OPCOA OPB815 £1.302N5777 50pTIL81 PHOTO TRANSISTOR ElTIL38 INFRA RED LED 5/E14N25, OP12252 OPTO ISOLATOR 50pPHOTO DIODE 50P 6/C2MEL12 (PHOTO DARLINGTON BASE n/c) 50pLED's RED 3 or 5mm 12/£1 100/£6LED's GREEN OR YELLOW 10/£1 100/£6FLASHING RED OR GREEN LED 5mm 50p 100/£40HIGH SPEED MEDIUM AREA PHOTODIODE RS651-995 £10 ea

STC NTC BEAD THERMISTORSG22 220R, G13 1K, G23 2K, G24 20K, G54 50K, G25 200K,RES 20 C DIRECTLY HEATED TYPE £1 eaFS22BW NTC BEAD INSIDE END OF 1" GLASS PROBE RES20-C 200R El eaA13 DIRECTLY HEATED BEAD THERMISTOR 1k res. ideal foraudio Wien Bridge Oscillator £2 ea

CERMET MULTI TURN PRESETS 3/4"1OR 20R 100R 200R 250R 500R 2K 2K2 2K5 5K 10K 47K 50K100K 200K 500K 2M 50p ea

IC SOCKETS14/16/18/20/24/28/40 -WAY DIL SKIS El per TUBE8 -WAY DIL SKITS £2 per TUBE32 -WAY TURNED PIN SKIS. 7k available 3 for £1SIMM SOCKET FOR 2 x30 way SIMMS £1

SOLID STATE RELAYS40A 250V AC SOLID STATE RELAYS £10

POLYESTER/POLYCARB CAPS330nF 10% 250V AC X2 RATED PHILIPS TYPE 330 £20/100100n, 220n 63V 5mm 20/£1 100/£31iV3n3/5n6/8n2/10n 1% 63V 10mm 100/E5101'05n/22433n/47n/66o lOmm rad 100/E3.50

100/£3100/E6 (El)

100/£1010/£1

50p ea100/£6

£1£1.25£2.50£4.00

£3£4£3

5/E1

5/£1 ra 22mm, 2p2 100V rad 15mm3/£1 lOrV33rV47n 250V AC x rated 15mm

£1 each lu 600V MIXED DIELECTRIC10/C1 1p0 100V rad 15mm, 1p0 22mm rad

50p F BITSEl ea AA ONI MICROWAVE DIODES TYPES DC2929, DC2962,

5/E1 4229F1/F2 El EAXTAL FILTERS 21M4 55M0 £2 ea

10/E1 ALL TRIMMERS 3 for 50pPCB PIN 200/£1TO.220 micas + bushes 10/50p 100/£2TO -3 micas + bushes 15/£1Large heat shrink sleeving pack £2IEC chassis plug filter 10A £3POTS SHORT SPINDLES 2K5 10K 25K 1M 2M5 4/E140k U/S TRANSDUCERS EX-EQPT NO DATALM335Z IOMV/degree CLM234Z CONST. CURRENT I.C.BNC TO 4MM BINDING POST SIM RS 455-961BUTTON CELLS SIM. AG10/AG12MIN PCB POWER RELAYS 10.5v COIL 6A CONTACTS 1 p oledo £1AVEL-LINDBERG MOULDED TRANSFORMER TYPE 061015+15V 10VA QTY. AVAILABLE £2 eaBANDOLIE RED COMPONENTS ASSORTED Rs, Cs, ZENERS

£5/1000LCD MODULE 16 CHAR. X 1 LINE (SIMILAR TO HITACHILM10) £5KYNAR WIRE WRAP WIRE £1/REELOPI1264A 10kV OPTO ISOLATOR E1.35 ea 100+ £1 ea'LOVE STORY' CLOCKWORK MUSICAL BOX MECHANISMMADE BY SANKYO £1 eaTelephone cable clips with hardened pins 500/E2

DIODES AND RECTIFIERSAl 15M 3A 600V FAST RECOVERY DIODE 4/E1

£1/prClCl£1

4/E1

VIOLET 5-105pFYELLOW 5-65pF RED 10-110pF GREY 5-25pFSMALL MULLARD 2 to 22pF 3 FOR 50p £10/100TRANSISTORS 2N4427, 2N3866, 2N5109 80p

ea.CERAMIC FILTERS 4M5/6M/9M/10M7 60p eaFEED THRU' CERAMIC CAPS 1000pF 10/E1SL610 E56 VOLT TELEDYNE RELAYS 2 POLE CHANGEOVER £2(BFY51 TRANSISTOR CAN SIZE)2N2222 METAL 5/£1P2N2222A PLASTIC 10/E12N2369A 5/£1

PLESSEY ICS EX -STOCKSL350G SL360G SL362C SL403D SL423ASL521B SL523C SL5416 SL850C SL1021ASP8655 SP8719DG

MONOLITHIC CERAMIC CAPACITORS10n 50V 2.5mm 100/£4.50100n 50V 2.5mm or 5mm 100/E6100n ax short leads 100/£3100n ax long leads 100/£5100n 50V dil package 0.3" rad 100/E81µF 50v 5mm E6/100

SEND £1 STAMPS FOR CURRENT IC +SEMI STOCK LIST - ALSO AVAILABLE ON31/2" FLOPPY DISK

MAIL ORDER ONLYMIN. CASH ORDER f5.00. OFFICIAL ORDERS WELCOME

UNIVERSITIES/COLLEGES/SCHOOLS/GOVT. DEPARTMENTSMIN. ACCOUNT ORDER £10.00

P&P AS SHOWN IN BRACKETS (HEAVY ITEMS) OTHERWISE 95p

ADD 171/2% VAT TO TOTALELECTRONIC COMPONENTS BOUGHT FOR CASH


AUDIO

Over the years,designers have battledwith the laws of physicsto eke more bass fromever more compactenclosures. Acousticdesign consultant DavidBerriman examines thecompromises involved intrying to squeeze outthat last decibel.

Ever since the earliest loudspeakers -literally loud -speaking telephones -engineers have tried to extend the

bandwidth of the audible spectrum at bothends. While high frequencies have been well -served by small and light diaphragms drivenby various means, low frequencies haveinvariably involved bulky cabinets and/or lowconversion efficiencies.

At lower frequencies, sound pressure level isdirectly related to the amount of air theloudspeaker diaphragm can move. Volumevelocity is the product of the radiatingelement's velocity and area: the higher thefigure, the greater the sound pressure. A largecone is therefore desirable but most peoplewant more bass from a smaller box

The obvious alternative is to make thesmaller diaphragm move further. Extendedtravel needs either a very long voice coil in ashort gap or a short coil in a long gap. Theformer wastes amplifier power while the latterrequires a much larger, more costly magnet.

Increasing cone area improves couplingbetween the cone and the air load, increasingefficiency. However the resulting cone is alsousually heavier, offsetting the sensitivity gain.

E BASSRACE

So large area diaphragms are usuallyrequired to produce deep, powerful bass.There is no easy way of overcoming thisgeneral rule, though many schemes have beendevised to achieve the best possible bass for agiven size of driver and/or box.

Woofers can be designed to suit specificloudspeaker systems, (or cabinets optimisedfor woofers) by using a calculator, Thiel Smallparameters and a few basic formulae. Farbetter are modern purpose -designed cadpackages such as Leap, which achieve thesame ends much faster and more accurately.

One thing is certain. Woofers need somesort of enclosure to work properly - and thereare plenty to choose from.

EnclosuresOpen baffle. At its most basic, this is a largeboard with the woofer bolted in a hole at thecentre. Unfortunately, the pressure wave fromthe rear of the woofer passes around the baffleand because it is 180° out of phase with thefrontal radiation it causes cancellation at lowerfrequencies. At high frequencies the wavelengthis shorter and cancellation does not occur.

Between the two frequency extremes there

is a rather uneven low -frequency roll off.Though the woofer on its own is basically asecond -order system (and thus rolls offasymptotically to 12dB per octave belowresonance) the roll -off for most real openbaffles starts well above this frequency andreduces the output by a half for every halvingin frequency, which is equal to 6dB peroctave. For a circular baffle of radius r, therear wave must travel 2r to reach the front. So,there is a roll -off in output at frequencieswhere the wavelength is longer than the bafflediameter (=c/baffle diameter where c is thespeed of sound in air).

Increasing baffle size lowers the turn -overfrequency, but giant baffles are domesticallyunacceptable. One solution is boosting theoutput at 6dB per octave below the 'knee', butthis dramatically increases the cone excursion.Boosting can be achieved by adding a largeinductor or by active compensation.

Bass from such a system can sound verynatural. Even so, open -baffle bass loudspeakersare limited to the esoteric end of the hi-fimarket and will only ever have a limitedfollowing. For popular consumption, mostsystems come in more compact packages.


AUDIO

Fig. 2. Acoustical circuit for a woofer in a sealed box (a).Mechanical impedances in Fig. 5 have been converted to acousticimpedances. Input is now a pressure source. In (b) is a simplifiedacoustic circuit for a woofer in a sealed box, combining woofer,box and leakage losses to form Ra. These total capacitancesrepresenting driver Cas and box compliances Cab to form Cat. This

L vc

C mes

Fig. 3. Electrical circuit for sealed box woofer system withLee representing woofer compliance and C, the wooferdiaphragm moving mass. Impedance peaks at resonancewith a sharpness dependent on losses.

Fig. 1. Mechanical/acoustical circuit for a woofer in a sealedbox. Rab and Rat represent box and leakage lossesrespectively. The cone is represented by a transformer ofratio 15d, where Sd is cone area, thus convertingmechanical force to pressure and velocity u to volumevelocity U. Input is force from the voice coil.

is the minimum form. Sound pressure is proportional to theproduct of U and frequency. Provided Q is low enough in thiscircuit, U will rise at 6dB per octave below resonance and fall at6dB per octave above. Sound pressure is proportional to theproduct of U and frequency, giving a flat output above resonanceand a high pass 12dB per octave filtering action below.

C mes

Fig. 4. Capacitive tuning of a sealed box extends bass byinteracting with Cat and C, while also forming a third orderhigh-pass electro/acoustic filter. Only the electrical equivalentsare shown here.

Sealed enclosure. For the simple sealed -boxloudspeaker , the acoustical compliance of airin the cabinet, Cab and the mechanicalcompliance of the woofer suspension Cms, Fig.1, behave acoustically like series capacitors.These are Cas for the speaker and Cab for thebox, as shown in the acoustical circuit, Fig. 2.When transferred to the electrical 'side' of themodel, Fig. 3, they become parallel inductorsof total value kat.

Conversely, the woofer's moving mass Mmsbehaves like an inductor on the acousticalside, but is transformed via the voice coil andcone to a capacitor Cmes on the electrical sideas in Fig. 3. The resulting inductance andcapacitance form a damped resonant circuit.Clearly, the presence of the box increases thestiffness on which the moving mass resonates,thus raising the resonance frequency.

From the acoustical viewpoint of theequivalent circuit, placing one capacitor,represented by the enclosure, in series withanother, the suspension, reduces thecapacitance and raises the resonancefrequency. From the electrical equivalentviewpoint this is like placing two inductors inparallel. Their total value is reduced whiletheir resonance frequency is increased.

Whichever way it is viewed, the end result isthe same.

With a small box, a very compliantsuspension is needed to keep the resonancefrequency low and assure a decent low -frequency output. If the BL product -representing the woofer's electrical/magneticsystem - is too high, the total electrical Q ofthe speaker (Qts) is too low and electricaldamping is too great. Efficiency is gained, butat the expense of bass loss. A BL productwhich is too small (higher Q,$) loosesefficiency and can cause a rise in bass outputnear and above resonance. In addition,transient response can become poor and bassmay tend to boom.

Final Q of the speaker in a closed cabinet isknown as Qtc. Generally in a sealed box, Qtcvalues above unity can tend to boom, whileQ values below 0.5 can sound very dry.Critical damping occurs at a Qte of 0.5, (-6dBsound pressure level at resonance). For the so-called Butterworth alignment (-3dB soundpressure at resonance) occurs when Qt, is 0.7.

Capacitance loading. A development of theacoustic suspension, sealed, or totally enclosedbox, is the use of a series capacitor to modify

the frequency response, Fig. 4. Serieselectrical capacitor C1 interacts with thecapacitive electrical equivalent of the woofermass Cmes and inductive electrical equivalentof the driver/air compliance kat to form anelectrical/acoustical filter. This modifies theinput current, and hence volume velocity, atlow frequencies.

The net result of a correctly aligned filter isthat the low -frequency roll -off changes fromsecond order to third order and the -3dBfrequency is reduced. This makes capacitancetuning a handy way of extending bass withoutincurring the poor sub -bass woofer loadingwhich a reflex port causes. Expressed anotherway, the loudspeaker can be more compactand protected from being over -driven.

Input to the woofer is reduced at very lowfrequencies, thus curtailing excursion andimproving power handling below resonance.However, a major disadvantage is thedeterioration in sound introduced by the large -value capacitor of a few hundred microfarads.Typically, for economic reasons this is anelectrolytic type, which is hardly ideal in termsof linearity.

Reflex. In reflex loading, a hole, or port,


AUDIO

Driver characteristicsAssuming no influence from output from the rear of thespeaker, sound output above resonance is flat. Below it fallsoff by 12dB/octave. Strength of the motor is determined byflux density and wire length in the gap. This is expressed asBxL in newtons or testa metres.

Voice coil and cone are interfaces between the acoustical,mechanical and electrical parts of the woofer. To analyseelectro/acoustic systems, acoustics engineers have turned toelectrical engineering and borrowed ac circuit analysismodels which have proved ideal

Acoustical parameters must first be converted into.electricalanalogues. The results of these transformations are circuitsare quite different from electrical circuits. They are notdirectly comparable to ac filters, but should be thought of aselectro/acoustic filters.

Due to interaction between motor/generator, mechanicaland acoustical parts, the loudspeaker's modulus ofimpedance curve shows a peak at low frequencies whereresonance occurs. The shape and centre frequency of thiscurve is determined not only by electrical parameters, butalso by the acoustical and electrical ones.

For more involved systems, such as reflex loaded woofers,the impedance curve is more complex. Normally it showstwo humps representing the higher and lower frequencyresonances either side of a dip at port resonance.

Until Thiel and Small's work in the early 1970s analysingdrive units was difficult. They derived Thiel/Small parameterswhich are now used by virtually all loudspeaker designers tohelp them align woofer systems.

Cone amplitude

-G°n

tm:*. (1"/".

--(at)

ac'cr:

Boa

eye

\

Cele';

SPL+12dB/oct.

fs or fc(resonance)

Cone amplitude, cone velocity and acceleration versus frequencyaround a speaker's resonance point. Above resonance the wooferdiaphragm is controlled by its own mass and the much smallermass reactance of the air load. With constant force from the voicecoil, acceleration is constant and amplitude drops at 12dB peroctave with increase in frequency. Cone velocity drops at 6dB peroctave as frequency rises. As frequency is reduced belowresonance, the cone becomes under control of the suspensionstiffness and the stiffness of air trapped in an enclosed box. Coneamplitude levels off, velocity drops at 6dB per octave andacceleration at 12dB per octave.

Rc

Electrical

e BI/ Sog- Rc + js)Lc

,c Voice coil

`t.r?'<lot,Pao

( a 142 )

FREQUENCY --b.

SPLcone acceleration

6cra./z.L'e .

(alit )

Sound pressure p at distance r is proportional to frequency and is

P=21

Clearly, P, follows the acceleration curve, giving rise to the classic12dB/octave high-pass filter curve for a woofer in a sealed box (orvery large baffle). This holds good provided the wavelength ofsound is greater than ft,. (or frequency is less than cAtr). At andabove this transition, radiation impedance on the diaphragmchanges from predominantly reactive to resistive. Sound dispersion

narrows and cones normally cease to behave as pure pistons, sothe neat theory starts to fail. For bass, the theory works well.

Mechanical 1/Rms

Za front

rear

Cone Air

Acoustic

Interfacing between an audio signal and air via a loudspeakervoice coil and cone, above. Models such as this are excellentloudspeaker design tools. Note that this is a mobility circuit, inwhich the parameters are the inverse of their impedances.Acoustic load on the diaphragm affects its ability to moveunder the influence of the mechanical force of the 'motor'created by the input current to the voice coil and magnetsystem. Working as a generator, acoustical qualities aretransformed to mechanical via diaphragm and then to electricalones via the voice coil. In impedance circuit, left, the variousparameters have been converted into their acousticequivalents. Here, the output we are concerned with is thevolume velocity U.


AUDIO

-30

0.1 0.2 0.5 1 2 5

NORMALISED FREQUENCY f/f

Cavity 1

volume VB1Driveunit

Cavity 2

volume V B2

Passiveradiator

Fig. 5. Outputs ofwoofer and vent in areflex loudspeaker.The vent peaks aswoofer outputdiminishes. In acorrectly -designedloudspeaker theseshould combine toform a smoothcurve with a forth -order high-passfilter shape.

10

Fig. 6. In a double -cavity, orcoupled -cavity loudspeaker,either a port or passive radiatorcouples the woofer to the air.Volume velocity from theport/passive radiator can behigher than that from the wooferalone, boosting sound outputover a band of frequencies.Unlike reflex speakers, thecoupled cavity has a bandpasscharacteristic.

located in one of the cabinet's panels,introduces another second -order resonantsystem to the simple second -order closed box.Air mass in the port behaves in the acousticalcircuit like another inductor, which resonateswith the enclosed air's acoustical capacitance.

Inductance in the acoustical circuitrepresenting the driver's moving mass, M- as,also resonates with this capacitance. In theelectrical equivalent, the roles of capacitanceand inductance are reversed. The upshot isthat the extra tuned circuit due to the port,coupled to that of the woofer and internal airvolume, further complicates the electricalequivalent circuit filter. The total soundpressure radiated is the vector sum of thatfrom the cone and the port. In other words,this is derived from the volume velocity of theport, added to the woofer volume velocity.

The two equivalent second -order tunedcircuits interact electro/mechanically andcombine acoustically to create fourth -orderhigh-pass filter, Fig. 5. In the reflex system,the single peak in the impedance curvepresented by the open backed and sealed boxloudspeakers is replaced by two peaks,representing upper and lower resonances.These are either side of a dip at the portresonance, where restricted cone motiongenerates minimal back EMF.

Acoustically, at the port resonance, conevelocity and excursion are at a minimum dueto extra loading by the high pressure of air inthe cabinet. This is produced by the air in theport, which rushes into the cabinet

simultaneously with the cone and at greatervelocity - an excellent example of resonanceamplitude exceeding the excitation. Theincrease in port output here fills the 'hole'created by the lack of cone output atresonance, thus transferring the main radiatingzone from diaphragm to port, Fig. 5.

In a well -designed reflex, the combinedresponse can be smooth down to the low -frequency roll -off. With a poor one, there areunwanted peaks and troughs and an inferiortransient response. In a well -designed reflexloudspeaker, power handling is improved andcone excursion reduced, at the same time asmoving more air at low frequencies.

For a given sensitivity, the reflexloudspeaker can be given a wider bandwidththan a comparable sealed box. Conversely, itcan have a greater sensitivity than a sealed boxhaving a comparable bandwidth. The maindrawback is that below the port resonance, thecone excursion rises rapidly. Because it isunrestrained by the box's air spring, the conemoves further than that of a comparableacoustic suspension design. This can beovercome by capacitive loading, as with thecapacitively -coupled sealed box, whichreduces the electrical input at low frequencies,while extending bass and creating a fifth -orderhigh-pass filter characteristic.

Passive Radiator Passive radiatorloudspeakers are a sub -division of the reflexcategory. The difference is that the port's airmass is replaced by a subsidiary drone cone

with its own suspension. Mass of this droneresonates with the air spring in the cabinet,and the compliance of its own suspension. Themain advantage of this type of bass loading isthat it avoids the sound of air chuffing throughthe reflex port.

Unfortunately, passive radiator loudspeakershave, in the past, been even more difficult todesign than good reflex loudspeakers, so thebreed has gained a rather poor reputation,perhaps unfairly.

Coupled cavity. This type of enclosure is across between a sealed enclosure and a reflex.Output from a conventional reflex loudspeakercomprises that of the port, driven by the rearof the cone, plus that from the front of thecone. In contrast, output from a coupled cavitycan be considered to be solely that from theport driven by one side of the cone. The otherside of the cone couples to a sealed box,which disposes of the sound from this side andacts merely like an acoustic spring, Fig. 6.

Figure 7 shows the acoustic equivalentcircuit for a coupled cavity using a passiveradiator. The coupled cavity is not a high-passfiltering reflex. It acts as a band-pass filterwith lower and upper slopes of 12dB peroctave. Though reflex in concept, it is nosharper acting than a sealed box and , intheory at least, achieves automatic low-passcrossover filtering.

In theory, no crossover filter is required.However, unlike the interior of sealed boxloudspeakers, the front cavity cannot becompletely filled with absorbent wadding.Consequently, air resonances at higherfrequencies, even when damped as far as ispracticable, can rather spoil the sound.

Advantages of coupled -cavity loading aremany. The main gains are a greatimprovement in the amount of air that can beshifted by the woofer at low frequencies andthe bandpass nature of the design. Because thecoupled cavity is a tuned system, the portoutput can exceed that of the cone alone,while cone movement is restricted. Thisreduces distortion and enables more power tobe accepted before the maximum linearexcursion is reached.

For sealed rear cavities, cone excursion islimited below resonance, which is good forpower handling. By altering the port andcavity dimensions, sensitivity can be tradedfor bandwidth, or vice verse. This makes it avery flexible system but one which reallyneeds cad.

Transmission line. This alternative is namedafter the transmission line of classicalelectrical theory with lumped absorption, massand compliance (lumped resistance inductanceand capacitance). In theory, it attenuates thewave from the cone rear, preventing it frombeing reflected at the open end to influence thewoofer, or radiating from the open end.

In reality, the transmission line could not bemore different. Often, the line is lightlydamped and tuned, rather like an rf co -axial`stub'. Sound takes a finite time to travel


AUDIO

down the line which is chosen to be a quarter,or sometimes an eighth of a wavelength ofsound in the line at frequencies where bassreinforcement is required.

The delay brings the output from the end ofthe line roughly opposite in phase to that fromthe rear of the cone. Thus it is more or less in -phase with the front of the cone over quite awide range of frequencies and reinforces thelow -frequency output of the loudspeaker ataround the woofer resonance and below. Inaddition, the reflected wave from the mis-terminated end travels back through the line toincrease the air load behind the cone.

Similarly to reflex loudspeakers, this reducesthe excursion and simultaneously reinforcesbass output, but over a wide range, not asharply tuned frequency band. Commercialtransmission lines thus behave a little like areflex loudspeakers, but not quite. They alsoadd their own pipe -like resonances.

Horn loudspeakers. While the previousexamples behave like relatively simple high-pass filters, the horn loudspeaker is rathermore complex, behaving also like an acoustictransformer. This is because the flare of thehorn couples the large air load at the mouth onto the smaller diameter of the drive unit end.The driver diaphragm is very much bettercoupled to the air than without the horn. Thusa relatively small diaphragm is almostmagically provided with a very large radiatingarea without the penalty of a large and heavydiaphragm.

The horn's acoustic impedance is largelyresistive, Fig. 8, and very high in value. As aresult the driver becomes resistance controlledover a wide bandwidth, instead of masscontrolled. The resistive load addsconsiderable damping to the diaphragm,which also reduces resonances within it.

As always there is a trade off - in this casesize. Below the horn's cut-off frequency, itsresistance falls away sharply, with a

corresponding rapid drop in output. At near tocut-off, with a finite -length horn, impedancecan fluctuate by quite a wide margin, thusgiving rise to large unwanted variations inoutput and corresponding audible colorations(honking). Careful design can help minimisethese variations, but they can never beeliminated.

Ideally, to accommodate the required flareprofile and work properly down to 30Hz, ahorn would need to be over 9m long. Inpractice, designers have brought cabinet sizesdown considerably by folding the horn (whichadds colouration) and using the room corneras an extension of the horn.

Quart from a pint pot?By now it should be apparent that deep bassmeans large woofers and cabinets. So, it isperhaps not surprising that designers havetried various ideas to extract deeper bass fromcabinets than would normally be possible. Inparticular, the 'Holy Grail' of deep bass fromvery small boxes has been with us for manyyears. Once the established techniques of

using a small long -throw, low sensitivitywoofer in a small cabinet and reflex orcapacitive loading have been investigated,other means must be sought to extend bass orshrink the box.

EqualisationThis involves boosting bass output at

frequencies where it would otherwise roll off.For instance, with a sealed box loudspeaker,having a Qt, (system Q at resonance) of 0.7,and -3dB at system resonance (the so-calledButterworth alignment), electrical boost of3dB could be applied at resonance, risingasymptotically to 12dB per octave as

frequency is reduced.At some low bass frequency the electronic

boost is arrested and changed to a low-passcharacteristic. Thus the frequency response islevelled down to a cut-off frequency lowerthan the system resonance, below which thebass rolls off at greater than 12dB per octave,due to the electronic filter.

Putting voice -coil thermal considerations

aside, the main limitation to power handling atlow frequencies is cone excursion. With anormal sealed box, or open baffle loudspeaker,cone excursion increases at 12dB per octaveas frequency is reduced towards resonance. Atresonance there is a transition: below thisfrequency the woofer diaphragm is understiffness control: and the increase in theexcursion levels off

However, while electronic equalisationboosts bass, it also increases cone excursionbelow resonance, placing much greaterdemands on the woofer and limiting lowfrequency power output. Even within thepower limit, distortion is bound to rise asfrequency drops.

With some manufacturers, the sub -

resonance boost is made programmedependent in an attempt to minimise theexcursion problem and the distortion ordamage which over -driving can cause. Inother words, bass boost is made greater forquiet bass sounds and smaller for loud basssounds. However, sudden deep bass notes can

1.2

R3 1.0

0.8

0.6

0.4

0.2

RAT MAS CAT

Wi ( UMP D

Fig. 7. Simplified acoustic analogue circuit for the bandpass, or double(coupled) cavity system. Input is pressure from the cone, output is volumevelocity from the port.

Exponential horn

Ro

PistonRp

40 100 200 400 1000 2000 4000 10,000

FREQUENCY Hz

Fig. 8. Acoustic resistance and reactance acting at the throat of a theoretical infiniteexponential horn, and of a piston in an infinite baffle. The resistive element for the horn ishigher and very a wide band, while reactance drops very low at high frequencies. A horn canload a diaphragm with a high resistive air load, reducing amplitude of movement whilecoupling the diaphragm more effectively to the air. This gives greater efficiency.


AUDIO

catch the system out and as a result they canoften be heard working.

Bass equalisation can just as easily beapplied to reflex loudspeakers. There is awhole family of bass alignment curves whichare well -suited for this application. Of course,this results in an even sharper high-pass filtercut-off than an equalised sealed box, witheven greater phase shifts. Such systems mustbe designed very carefully if bass quality isnot to suffer.

Negative Output Impedance. Anotherapproach is to make the power amplifierbehave as part of the loudspeaker. One way ofdoing this is to use negative feedback to createa negative output impedance, cancelling outthe voice coil dc resistance. This enables theloudspeaker driver and cabinet to be designedas if it had no dc resistance losses at all. Witha reflex loudspeaker, this allows the Q of theport resonance to be made higher, becausethere is less waste of energy through damping.

Though the use of negative outputimpedance was used to improve woofercontrol as long ago as 1950s, this idea wasresurrected a few year's ago by Yamaha, wholaunched Active Servo Technology, which isstill used in some of their equipment . The

main disadvantage of this technique is that thepower amplifier and loud speaker must bedesigned for each other, so it is hardly auniversal solution. However, it does have theadvantage that standard bass units can be used,which minimises the cost increase.

Motional feedbackNone of the above bass systems requiresanything other than standard drive units fortheir operation. However, the only way tocontrol the woofer absolutely is to include it inthe feedback loop in a totally active system.To achieve this, a transducer must be used todetect cone motion - either position,acceleration or velocity. Output from thetransducer is modified and used, in classiccontrol -system fashion, to derive an errorsignal for a feedback amplifier.

The extra complexity and cost of suchsystems has restricted motional -feedbackwoofer loudspeakers in the past to mainlystudio applications. A variety of sensingsystems may be used, from capacitivedisplacement sensors, to accelerometers.Alternatively the signal from an extra windingon the voice coil can be electronicallyconverted to displacement or acceleration.

Not only can motional feedback extend bass

response below the normal cut-off, it can alsogreatly reduce non-linear distortions -important if the voice coil and cone aremoving a long distance. Woofer distortionalways rises at low frequencies, so motionalfeedback can help solve one of the mainproblems in bass reproduction.

Despite numerous 'come backs', motionalfeedback woofer systems seem to havereceded in recent years, though Tannoy hasrecently announced a new system. Perhapswith today's digital technology it could be duefor a revival. Certainly the extra signalprocessing required could be included on achip with relative ease for use within aloudspeaker providing on -board dacs, digitalcrossover filters and internal power amplifiers.Similar so-called digital loudspeakers havealready been introduced by Philips andMeridian.

Over the years we may see new attempts toget deeper bass from smaller boxes. Most ofthese 'new' ideas are likely to have alreadybeen tried in one form of another, though thetechnology used will undoubtedly beadvanced. Ultimately designers are up againstthe same laws of physics as their predecessors,who so many times used those same laws totheir advantage.

FREE TO SUBSCRIBERSElectronics World offers you the chance to advertise

ABSOLUTELY FREE OF CHARGE!Simply write your ad in the form below, using one word per box, up to a maximum of twenty

words (remember to include your telephone number as one word). You must include your latestmailing label with your form, as this free offer applies to private subscribers only. Your ad will

be placed in the first available issue.This offer applies to private sales of electrical and electronic equipment only.

Trade advertisers should call Pat Bunce on 081-652 8339All adverts will be placed as soon as possible. However, we are unable to guarantee Insernon dates. We regret Ince we are uncluie to enter into correspondence

with readers using this service, we also eserve the right to reject adverts which do not fulfil the terms of this offer.

Please send your completed forms to:Free Classified Offer: Electronics World, 11th Floor, Quadrant House, The Quadrant, Sutton,Surrey SM2 5AS


RF ENGINEERING

Sloping vees forVHF and UHFArmed with performancefigures, Richard Formatocontends that the slopingvee antenna is unjustlyignored forhigh -frequency use.

RF design softwareIf you want to experiment withVHF/UHF vees, a sloping veemodelling program is essential.It is the only way to investigatetrade-offs as various antenna orground parameters arechanged. Radiation patterns inthis article were computedusing inexpensive PC compati-ble antenna design softwarefrom Phadean Engineering CoInc. The company is at PO Box611, Shrewsbury MA 01545-8611 (SAE for prices).

WireRadiatingElements

Hf

When antennas for VHF and lowerUHF band are mentioned, mostengineers think of whips. Yagis,

log-periodics. The sloping vee is conspicu-ously absent from this list. But the vee hap-pens to be a superb antenna - especially atshorter wavelengths. Yet it isn't well knownor widely used by amateur operators on thehigher frequencies.

The sloping vee is inexpensive, mechani-cally and eectrically simple and easily trans-ported and installed. Most importantly, it pro-vides excellent gain -bandwidth performance -particularly for single -band operation. Becausethe radiating elements arc inclined wires, veesalso provide the added bonus of inherentpolarization diversity.

A sloping vee comprises two radiating wiresdiverging from the antenna feed point, Fig. 1.A typical installation is shown in Fig. 2. Notethat Fig. I is a perspective view in which bothresistors R are at the same height H1. Thesenon -inductive resistors terminate the radiatingelements. The resistance value is half theantenna input resistance, and the power ratingis typically 10 to 20% of the maximum anten-na input power.

Incident energy that has not been radiatedinto space is absorbed by the resistors. Thissuppresses reflections which would otherwisegenerate standing waves and create strong res-onances. The resistors are connected by ashorting wire to complete the current path.

Fig. 1. Slioping vee antennas provide excellentgain -bandwidth performance, particularlywhen used on a single -band.

Incident Wave

R -

Reflected Wave

Ground Level

R

ShortingWire

Since the sloping vee is a balanced radiatingsystem any unbalanced feed line, such ascoaxial cable, requires a balun. The balunshould have the lowest possible insertion lossand flattest possible response over the vee'soperating frequency range. As an engineeringrule -of -thumb, the vee is considered a 600Qantenna. To match to a 50Q feed system, abalun with a 12:1 impedance ratio is required(square of the turns ratio). In practice, the-6004" antenna may actually turn out to be a40052 or even a 90052 system, which, ofcourse, changes the balun requirements.

To give you an idea of how good a vee canbe, the three plots show computed verticalradiation patterns for a 6m sloping vee.Although not optimised, the antenna performsvery well. The design frequency range is the6m amateur band (50 to 54MHz). It is

assumed that the antenna will be deployedover average ground with an electrical con-ductivity of 0.002S/m and a relative permit-tivity (dielectric constant) of 8.

Antenna parameters are as follows.Diameter of the radiating element is 3.2mmwhile the apex angle - the angle betweenwires at the feed point - is 15°. Feed pointheight above ground is 6m and above groundterminating resistor height is 8m.

Input resistances of the vee are 455, 446 and437Q at frequencies of 48, 52 and 56MHzrespectively. Taking the average value of44652 as representative, each terminating resis-tor should have a value of 223Q. In practice,200 or 25052 is close enough. Since computedinput resistance varies only 4% between 48and 56MHz, this design should provide essen-tially flat VSWR from 50 to 54MHz.

The plots show total power gain in dBi(decibels relative to an isotropic radiator). Toconvert to dBd (decibels relative to a dipole ),subtract 2.15 since gain of a half -wave dipolein free space is 2.I5dBi. Note that the totalpower gain includes both horizontal and ver-tical radiated fields, as well as antenna radia-tion efficiency. Patterns were computed at 48,52 and 56MHz for three radiating elementlengths of 20, 40 and 60m as annotated on thecurves. These radiation patterns are in a verti-cal plane bisecting the elements (zero azimuthangle). They are plotted on linear scales whichprovide a more detailed view than polar plots.

Key computed performance parameters are


RF ENGINEERING

20

15

10

5

0

-5

10

Coax feed cable

/Feed point

Radiating element

Dielectric mastHeight, Ht

Fig. 2. Mechanically and electrically simple, the slopingvee antenna is easy to transport and install. It is

also inexpensive - particularly if you use atree for the mast. The dimensions can

be scaled for other higher (andlower) frequencies.

Termination

Length. LDiameter, D

go

Termination

Zero azimuth line

Earth surface

The following table summarizes key computed performance parameters:

L

LjGmax

(dBi)

Angle

(deg)

-3dB BW

tOsal

1st SL

fdBij

1st SL

(dBi/Grr axFrequency = 48MHz

20

40

60

7.7

13.3

16.3

12

11

11

12.6

12.0

10.6

0.5

-2.3

-1.1

7.2

15.6

17.4Frequency = 52MHz

20

40

60

8.5

14.2

17.2

11

10

10

11.9

11.0

9.9

1.9

-1.5

0.3

6.6

15.7

16.9Frequency = 56MHz

20

40

60

9.3

15.0

18.0

10

10

9

11.0

10.4

9.5

3.3

-0.9

1.8

6.0

15.9

16.2

summarised in the table. It is evident that thissimple antenna exhibits exceptionally goodperformance. With the largest element 160mlong, main lobe gain varies from 16.3 to18dBi between 48 and 56MHz. Maximumgain for all lengths occurs at take -off angles

This curve family shows antenna gain (dBi)versus take -off angle for three lengths over threefrequencies. It assumes the optimum case wherethe feed point at the apex is lower than thetermination side. The more convenientarrangement shown in the drawings may incur again penalty of up to 3dB.

Sloping Vee 48 Mhz Pattern5=.002, EpsR=8, D=.32cm, Hf=6m, Ht=8m

.- L=60 m

L=40 m\

L=20I

m

\ /' '

t l',,,,,,10 20 30 40 50 60

Take off angle (deg)

between 9 and 12°, which is a suitable rangefor long-distance links. Take -off angle can becontrolled by adjusting the radiating elementlengths and feed point and terminationheights. As expected, the shortest element of20m provides the lowest gain, but even its per-formance is very respectable (7.7 to 9.3dBi).

This design example shows how well thesloping vee performs at VHF and UHF. Asthe example illustrates, the physical size of ahigh gain vee can be large. But its dimensionsare not so imposing, compared to the size of aYagi providing the same gain. Of course athigher frequencies, the shorter wavelengths

Sloping Vee 52 Mhz Pattern5=.002, EpsR=8, D=.32cm,Hf=6m, Ht=8m

20

15

10

0

-5

10

Specialist antenna components

Non -inductive film power resistors fortermination: Power Film Systems Inc,Yellville, AR 72687.

Stranded 7x9 phosphor -bronze cable(avoids kinks and tangles): AstroIndustries Inc, Dayton OH 43432.

Strong non-metallic masting: J. T.Ryerson Co., PO Box 1111, Boston MA02103.

Toroidal ferrite balun cores: Radio KitInc, PO Box 97, Pelham NH 0076.

result in much smaller designs.Another advantage provided by the vee is

simple installation. The three different sizeantennas in the design example could bedeployed in a variety of places, for example,between trees, or hung from a building orother structure. The range of possibilities islimited only by your imagination. Most anten-nas do not provide the installation flexibilitythat the vee does. About the only caveat tobear in mind is that, like any antenna, thevee's performance is influenced by nearbymetallic structures. If they are too close to theradiating elements, parasitic effects maybecome a problem.

The vee's electrical performance is the samewhether an exotic stranded cable or a plainsingle -conductor wire is used. The main dif-ference is convenience. As far as masts go,trees provide the same results as fancy dielec-tric ones, with somewhat less convenienceperhaps, but probably more fun.

It is evident that the sloping vee antennaexhibits exceptionally good performance,despite its simplicity. Maximum gain for alllengths occurs at take -off angles between 9and 12° which is a suitable range for long-distance links

Radiating element length is L and main lobemaximum gain is Gmax. Take -off angle formaximum gain is given in degrees above thehorizon while approximate main -lobe beamwidth is in degrees between points 3dB downfrom maximum gain. First side -lobe level isgiven in dBi and relative to the maximum gain(decibels down from the main lobe).

7-1-..,L=60L.,440 m

m

, L=20

IY

I

m

I,r, ,

,\1, , ,\A

\../ A ,

70 80 90 0 10 20 30 40 50 60

Take -off Angle (deg)

70 80

20

15

10

Sloping Vee 56 Mhz Pattern9,002, EpsR=8,D=.32cm, Ht=6m, Ht=8m

10 20 30 40 50 60

Take -off angle (deg)

70 80 90

1 24 ELECTRONICS WORLD + WIRELESS WORLD February 1994

Think of studentsI was very interested to read theletter about your review ofElectronic Workbench (EW+WW,December 1993), and found myselffirmly in agreement with virtuallyall of it. But I feel Reg Williamson'sfavourable bias toward the programshould perhaps be balanced by astudent's view.

A student rarely has months tobecome familiar with a softwarepackage, a program like Workbenchbeing used for maybe one or twounits of a course, and here lies themain problem.

Workbench requires the dexterityof a watchmaker to placeconnections, and there seem to be afew inconsistencies in the way thecircuits operate in the package. Forexample, in the digital version oneof two identical circuits failed tofunction when macroed. Couple thiswith the fact that the help and printfacility do not always produce thedesired results, and perhaps it ispossible to appreciate that thispackage takes a considerableamount of time to master.

I have spent around three hourstrying to find a problem in aWorkbench circuit that I could havebuilt trouble free in one hour. Howmany students can afford this time?

At previous educationalestablishments I have attended, therewas a very restricted access toanything other than word processingsoftware, and even now I cannotguarantee to get any extra time witha technical package outside normalclasses.

Should I wish to investigate atypical one -to -twenty active deviceeducational circuit, one could befabricated from a collection of TTLchips or a few small signalsemiconductors fairly easily andquite cheaply. This can even be doneand tested at home if the interest,inclination, and a few pounds worthof basic test equipment are there.

To prototype the same circuitusing Workbench, I have to eitherensure that the circuit is completed,tested and all results obtained within1 to 2 hours, or try to arrange extralaboratory sessions that will notclash with the timetables.

I have been on the receiving end ofPC oriented courses for a few yearsand, in general, after gettingacquainted with syntax, filehandling, and other little quirks,have found the PC quite an aid todesign and development. Throw inthe odd bug (nearly every packagehas them), add a few badly writtenbatch files, courtesy of lecturers'and technicians' menu writing skills,restrict the time available with thetechnology, and the student ends upfrustrated, angry and, towards the

term end when the obligatorykilogram of printouts are due in,panic stricken.

Perhaps colleges and universitiesmight like to bear in mind that theaverage student cannot afford thecost of a package such asWorkbench. Unless theycontemplate computer access overand above normal courserequirements, a more practicalapproach to digital design wouldperhaps be more advantageous.

For analogue work I see no reasonto consider Workbench at all, asthere is a limited node version ofPspice and Probe freely availablefor students' own use.

I have spent quite some time withWorkbench and, like MrWilliamson, I consider the packagehas some very good points. Thetrouble is it takes a fair few mousemiles to appreciate them.H. NearyStoke on Trent, Staffs

Solid airDick Manton was quite right topoint out (EW + WW, November1993) for the benefit of those whomight have been confused that mycomments on coaxial cable referredto solid dielectric types. Air spacedtypes offer higher velocity ratios,approaching unity. For example,Andrew Heliax air dielectric coaxialcable type HJ4-50 (0.5in corrugatedsolid copper outer, air -spaced innersupported on a helical spacer cord)has a 50Q impedance and a velocityof 91.4% of c.

Feeding these into the formula inmy article. L (per metre)=Zoxb gives152nH/m for this cable, well belowthe free space value of 1256nH/ni.

Likewise, I would imagine thatC/m for an air spaced 1200Qbalanced line is well below the freespace value. As for velocity ofpropagation in a line, coaxial orotherwise, this can vary widely.

Silent problemsFive recent papers on distortion in power amplifiersby Douglas Self make very interesting reading and itis good to have the various aspects of the subjectconsidered as an entity. They will be recommendedreading for my final year undergraduate audiosystems engineering class.

He stresses, quite correctly, the need to avoid thegeneration of high order harmonic andintermodulation products, but appears to accept thatfor stability of the feedback loop, the high -frequencyopen -loop gain of an amplifier should fall at-6dB/oct, starting from a frequency well within theaudio range.

Do not overlook that audible intermodulationproducts can arise from spurious signals withfrequencies above human hearing. The linearisingbenefits of negative feedback are therefore reducedjust where they are really wanted.

He also sees little advantage in using fets as outputdevices as opposed to bipolar transistors. I agree theyare not perfect devices and are in some respectsinferior to bipolars, but when used in suitable circuitsthey offer the possibility of considerably widerbandwidth output stages. This is their most significantadvantage.

In this case it is possible to employ a considerableamount of feedback and have an open -loopbandwidth covering the whole audio range. The highfrequency roll -off needs careful design to have anaverage rate of about -9dB/oct. This can be achievedby a succession of poles and zeros in the response,rather than one dominant time constant.

The immediate reaction of listeners to my amplifiersdesigned this way is that they have an enhanced trebleresponse. Further listening, and comparison withother types, alters this assessment to one of greaterclarity and realism, creating the illusion of moretreble. A stereo pair with suitable speakers arecapable of excellent imaging.Ivor BrownUniversity of West LondonUxbridge

LETTERSIn a coaxial delay cable as used for

delaying an oscilloscope's Yamplifier signal, a spiral woundinner is used. This reduces thevelocity of propagation way belowthe 0.65c typical of a solid dielectriccoaxial cable. In a loaded telephoneline the velocity of propagation maybe as little as a 20th of c.

However, when I said the valuesof L and C per metre in free spaceare the lowest that can ever beachieved I meant just that - in freespace. If a space wave ispropagating in a medium where thevalue of C is greater than88.5-12/m, that is in a dielectricmedium, it will be slowed down.This effect is used in certain types ofradar lenses.

Similarly, a wave propagating in amedium where L is greater than4xpx10-7/m will likewise be slowed,an effect used in certain waveguidestructures.Further to my comments in the

Douglas Self replies:"I thank Mr Brown for his comments. It is quite truethat in general I expect the open loop gain of anamplifier ultimately decrease at 6dB/octave, though Ido not assume that this rolloff begins "well inside theaudio range". The actual frequency at which rolloffstarts, as far as I can see, of very little importance initself; what matters is the amount of feedbackavailable at the upper end of the audio band, becauseonly here will the distortion of a Blameless amplifierbe measurable, and also the amount and slope ofrolloff at the unity loop gain frequency, because thissets HF stability.As I showed in the article on the voltage amplifierstage, with the aid of a cunningly placed resistor, youcan make the start of the rolloff occur at almost anyfrequency you like (20Hz if you wish) withoutaffecting the most critical part of the open loop gaincharacteristic. This is why I prefer to quote NFBfactors at 20kHz.I do indeed take a rather pessimistic view of fets inoutput stages, and have always found the greaterbandwidth to be more of a hindrance than a help.They may promise a higher bandwidth output stage,but do they deliver, especially when capacitivelyloaded? If Mr Brown has some data on this, I hope hewill share it with us. The worst drawback of fets isthat they are so depressingly nonlinear, despite whatyou sometimes read in the hi-fi press.I am well aware that there are other kinds ofcompensation, as future parts of the series will show.However, 1 don't much care for the 9dB/octaveapproach, because if this slope is maintained up to theunity -gain frequency, it directly reduces the stabilitymargins. It also requires a series of alternate poles andzeros that are not easy to fit into the conventionalamplifier topology I have been discussing. I think abetter method uses two pole compensation because: itis cheap and easy to implement if you know andavoid the snags; gives a stunning reduction in HFdistortion; allows the gain slope to be returned to6dB/octave before the unity gain frequency is reachedgiving HF stability no worse than standard dominantpole compensation."


LETTERS

article concerning a per -unitgenerator that always supplies a totalof 2W, wherever it is dissipated andwhatever the load, no one has so farwritten in to quote an earlierreference. While they are admittedlyacademic constructs rather thanuseful circuit arrangements, perhapsthe Hickman generators types I andII really are novel after all.Ian HickmanWaterlooville, Hants

Theories of scienceJudging from the sort of argumentspresented in these columns over thepast year or so, the jolt A. M. New(EW+WW, October 1993) gave uswas much needed. In the simplest oflanguage he reminded us of a truththat most of us seem to want toforget - all scientific theories arefiction.

This puts into perspective suchdisingenuous questions as how wecan accept theory based on theconstancy of the speed of light inview of the experimental evidencethat it is bent by gravity.

Those who propose such questionsseem entirely blind to the fact thatthey may be applied to othertheories. We might equally ask how,in view of the experimental evidencethat neither mass nor time areinvariants, we can possibly accept atheory based on the assumption thatthey are. And yet we accept andstrongly defend such a theory on adaily basis.

As far as the question aboutspecial relativity is concerned,Einstein dealt with the matter in1916, long before any experimentalevidence existed. In his bookRelativity, the special and thegeneral theory he points out thatgeneral theory predicts the bendingof light, which must call theassumption of the constancy of itsvelocity into question. and examineswhat this means for his earliertheory. He says, "We can onlyconclude that the special theory ofrelativity cannot claim an unlimiteddomain of validity: its results holdgood only so long as we are able todisregard the influences ofgravitational fields on thephenomena. for example of light."

If necessary, we might wellparaphrase this and say. "We canonly conclude that Newtoniantheory cannot claim an unlimiteddomain of validity. Its results holdgood only so long as we are able todisregard the influences of relativemotion on the phenomena, forexample mass and time."To fail to do this is to hold doublestandards, a failing that seems notuncommon even among physicists.The harsh fact is that all theories

Radio ham baloneyWhile I endorse most of the remarks contained in youreditorial comment (EW+WW, December 1993), I musttake issue on several of your opening points.

First, you quite rightly state that Tony Hancock'sradio amateur was "pompous, petty and technicallyincompetent", but then you go on to say that thisinterpretation "encapsulates the truth about the hobby".I strongly object to being described by any of theseadjectives and would think the majority of other radioamateurs agree with me, particularly when the remarksare made by a fellow amateur who edits a respectedtechnical periodical.

Secondly, it is open to debate whether "everyone...enjoys the Hancock sketch." The fact that many peoplefound the sketch offensive and insulting was evidencedby letters to editors of various magazines at the time,which I suggest was long before you obtained yourlicence.

Hancock was, in fact, demonstrating the "selfconscious and inane chatter" you attribute to the newinflux of radio amateurs, but unlike them wasbroadcasting to a far wider audience. Mr and MrsAverage are unfamiliar with amateur radio andtherefore assume Hancock's parody mirrors the truth.On the other hand, the trivial prattle heard on manyamateur bands is heard only by other radio amateurs,most of whom are capable of treating it with thecontempt it deserves.

While I do not condone 'cheque -book engineering'and am unhappy with many of the evolutionarychanges that have affected amateur radio over theyears. I am a realist and accept that times change.However much we might like to believe it, thelikelihood of a radio amateur working in his gardenshed coming up with an idea that will revolutioniseradio communication is fatuous nonsense.

have shortcomings by their verynature. Some may be shorter thanothers.We ought to be spending ourenergy on determining what thoseare and how they influence thedomain of validity of any particulartheory, not on arguing whether anyone theory is the truth. Truth in thiscase is - dare 1 say it - entirelyrelative?Alan WatsonMallorca, Spain

Not going round incirclesIt has been repeatedly demonstratedby men such as Poor, Ives, Dingle,Marinov, Beckmann, Sachs andHayden that Einstein's relativityinvolves circular logic, and is quiteimpossible. The idea of theconstancy of the speed of light in alldirections, independent of sourceand observer, can be shown toviolate the second law ofthermodynamics.

Many alleged conclusive proofs ofrelativity have been shown to bebased on fudged data, such asEddington's on the gravitationalbending of light, the perihelionprecession of Mercury, the Naystar

We, as radio amateurs, must not fall into the trap oftaking ourselves too seriously and thus forget thatamateur radio is a hobby to be enjoyed and not somevast army of latterday Maxwells or Heavisides.

To think otherwise is as pompous as Hancock'sportrayal of a typical radio amateur, but then you thinkthis encapsulates the truth anyway. Perhaps I havebeen wasting my leisure time over the last 34 years ontie air enjoying my hobby when I should have beenengaged in more productive pursuits whereby thefrontiers of knowledge could have been pushed furtherforward.

As a chartered engineer. I am content in theknowledge that I have contributed to the advance ofknowledge during the working day and see nonecessity to carry on in the same vein while pursuingmy hobby. We should concentrate our efforts intoimproving the image of the hobby rather thanhankering after passed glories that are unlikely to berepeated in the late 20th century.

Far more could be gained by cleaning up thelanguage and improving operating procedures on ourhard won frequencies than will ever be achieved bybemoaning the fact that revolutionary advances inspread -spectrum techniques are not being developed bythe amateur fraternity.A. C. Wadsworth G3NPFliorsham, West Sussex

You make some excellent points in you editorial onanateur radio (December 1993). However you fail tomention the fact that amateur radio is a hobby and thatin itself justifies its existence. Just as people still climbEverest, we can all experience the thrill of personalfirsts long after the achievement has becomecommonplace.G. P. Stanley G3MCKStain Middlesex

project, and Hafele and Keating's ontime dilation. Relativity is riddledwith anomalies and paradoxes: theEhrenfest paradox, the paradox ofself reference, the curvature of spaceparadox, the ruler paradox, the clockparadox, the simultaneity paradoxetc.To this day, all Michelson-Morley type experiments are unableto detect even a smidgen of theearth's purported 67,000mile/htranslational velocity around the sun.As Professor Lincoln Barnett said,"We cannot feel our motion -throughspace; indeed no experiment hasever shown that the earth actually isin motion".

Relativity is refuted by theaberration of starlight, the everydayoperation of numerous electro-optical engineering devices, theSagnac effect, and the results of theMichelson -Gale experiment, whichare all exactly what one wouldexpect in a geocentric universe.

There is much more recentevidence from astronomy andquantum physics to show that theearth is indeed the preferred frameof reference in the universe, locatedat or near its centre.

Bertrand Russel said, "Whetherthe earth rotates once a day fromwest to east as Copernicus taught, or

the heavens revolve once a day fromeast to west as his predecessors held,the observable phenomena will beexactly the same: a metaphysicalassumption has to be made".

See Geocentricity, Gerardus Bouw1993, "The earth is not moving -400 years of deception exposed".Marshall Hall 1992. and Thecosmos, Einstein and the truth,Walter van der Ramp. 1993.Amnon GoldbergLondon

Black hole errorIn in article on gravity andelectric force in a black hole(EW+WW, February 1993) asmall printing error managed tocreep through. The symbol h ofthe last equation was misprinted.The correct equation is:

G = c5a2 (2- a)2 (e / 4/1.2)4 / trh

D. Di MarioMilan, Italy


16 bits of difference?Over the five years since its launch, Mathcad has become faster andevolved from DOS to Windows. But according to Allen Brown itremains quirky and has progressed little in real problem solving power.

One of the foremost equation processing packages -Mathcad from Mathsoft - has been upgraded toversion four. The principal difference between this

version and its predecessor version 3.0 is its ability to takefull advantage of the 32bit architecture of the 386/486 PC.Surprisingly, 32bit 386 PCs have been available for atleast five years, yet very little software is actually coded in32 bits. Most PC software is still coded in 16bit formatwhich is a waste of the extended data bus width.

Installation of Mathcad 4.0 is straightforward. The onlynotable point is the creation of a sub -directory ofWINDOWS/SYSTEM called WIN32S. It contains thedynamic link libraries (DLLs) used in the 32bitprocessing.

Mathcad is a mathematical scratch pad allowing you toexpress equations directly on screen. Solutions can bedisplayed in tabular form or graphically in either two orthree -dimensions. Additionally, equations can be solved togive either numerical or analytical solutions using theMaple equation processor. In fact Mathcad can beenvisaged as a sort of super calculator.

Version 4 for Windows is installed like any otherWindows software and possesses the general familiarfeatures, including buttons and dialogue boxes. However itdoes require the virtual memory facility for swappingbetween hard disk and system ram. As the illustrationsshow, the screen display is rich in icons, permitting easyaccess of Mathcad's many features. The Windows versionalso allows you to use the range of Windows fonts forcreating text alongside equations, tables, graphs ordiagrams.

For modelling linear systems, Mathcad is a veryattractive tool. It can be effective for designing infinite -impulse response (TIR) digital filters, Fig. 1, since they arederived from a linear system model. The same applies tofinite impulse response (FIR) filters based on the windowdesign - again a linear system model. Things get a littlemore awkward when dealing with coupled linear systemssince one has resort to using Mathcad's matrix facilities.

Within Mathcad is a large array of easily accessible builtin functions. For example if you want to determine the

PCENGINEERING

1:1 Mathcad- pIR.MCD)File Edit Text Math Graphi :s Symbolic Window Hooks H I )

..

u [DEAcalaii ,eixiii3imi 9.ig. Eci t- v virl Q.luf°1:= Default TPY, III LAme, 10 til B11111 j ab ob alb

Plot of impulse response: Coefficits

ci,

..---..

....-....,

.......

7r,,a:...

III..k......rlxi

X.`.....x,

I

iOARYI

0.319071 I I I 8,832 1C":"

-0.00.

b'- ...

0.01£

-0.10!oIn 0445

b. -.I I T _ _ -1383,.,....,,_.3.25:

- 5 8V -100511 i I i i 0.094 ^"£,.0

0 ' 50 -8.4E8

6 51*

-33n, I

.1 1 1.±..

Pa i e 4 I I auto

condition for an optical fibre to support single mode thenyou need the first root of the equation,

J0(x)=Owhere J0 is a Bessel function. Figure 2 shows a plot of thisequation and from the plot the function is zero when x isaround 2. By using the ROOT function with your guess of 2,Mathcad will give the correct value of x=2.404.

Symbolic CalculatorOne of the attractive features of Mathcad is SymbolicCalculator. Developed by Waterloo Maple Software ofOntario, this is a derivative of Maple which performsalgebra and calculus analytically or symbolically.

Given an integral, you no longer have to work it out.Figure 3 shows an integration and an expansion of aseries. Even when manipulating matrices, variables can be

Fig. 1. Mathcadcan be quiteeffective atmodelling infiniteimpulse responsefilters.


PC ENGINEERING

Mathcad - [Untitled:21w File Edit Text Math Graphics Symbolic Window Boo rs Help

CIMEIM 41xlibleNIFT-1 clogDefault Text LtilAual Id 110 1.1Lnij-juiehiFinding the first root to the equation J0(x)0.

=0,1 15

The first ropt is close to 2.let

x =2

10(e) o root(50(x),x) =2.404

1,4 The root fiction finds a morexv accurate Iva_ue.X. 0 3 41 9 12 15

l'age 1 auto

lig. 2. there is a wide choice of built-in functions. This is an example of finding rootsto aid determining the condition for an optical fibre to support single mode.

.w. Mathcad - [Untitled:IIw File Edit Text Math GraphicsSymbolic Window Bookr. Help ....

3 coffin A xitiersi NV] aill=11 QIIiii1Plidgiklia Detatir Tevt fl lAnal II la d i d tobloblabj13

---*7

2 Performing an integral on.1

ds

4

8

c

rl",7,:,..

--k-r-

v,..

110(1

6 1

+

11 +

[14-c(ef-91

exp

Ix )

x

a

JJ

(E f- S 2))

rx2 4.x

2a

4- s2 - 111(1 + exp (s f - 61)) - s --

Performing a series expanson

3 4x

a3 aa_ 4.7n itPa

,.e 1 I auto

Fig. 3. 514111)01i( Calculator in action - integration and expansion of a series.

Mathcad - IRIC.MCDIfile Edit Text Math Graphics Symbolic Window Books Help

BOOa

-'Enter admittances between each node Fuld any higher -numbered

Constants 11 in,e, New Roman Finn

0 0 01

10] 010.

gicejslto"

21:101 0./10"

so

0 0

00 0 0 0

ri-qt. Tel

20.3 01-1

(006

0..5 0

10-T

Pa e 1 auto

Fig. 4. An example of network analysis with Mathcad. When manipulating matrices,variables can be used as elements in the matrix.

used as elements in the matrix - handy when you areperforming small scale network analysis, Fig. 4.

Using Symbolic Calculator for evaluating analyticalexpressions you also have the option of mixing constantswith numbers. For example if you put an actual value onthe upper limit of an integral sign and a variable on thelower limit, Mathcad will perform the integration, carryingthe variable through. With the previous version ofMathcad, Symbolic Calculator was rather slow but version4 makes it noticeably faster.

SmartMathAn interesting variation on the Symbolic Calculator isSmartMath which uses an expert system to best combinesymbolic and numerical evaluations. The ordinarySymbolic Calculator dutifully takes every expressionliterally and works out the answer. SmartMath on the otherhand examines what you have entered and works out theoptimum method of solution according to a set of expertsystem rules. It is difficult to divine why Mathcad 4 shouldcarry the Symbolic Calculator at all since SmartMathappears to do the same job in a fraction of the time.

Graph PlottingEven with the first version of Mathcad, graph plotting wasparticularly easy to perform. This was especially true fortwo dimensional graphs. Later versions of the product hadthe provision for drawing 3D surface plots but it was notparticularly easy to generate these since the user wasexpected to create a matrix of the surface first.

The same awkward and unappealing technique is used inversion 4. However the new version can produce polarplots with the same ease as two dimensional plots and isuseful for estimating field patterns from radiators, Fig. 5.In addition there is a new facility for contour plots whichhas contour labelling. This is common feature withcommercial graphics packages and it is reassuring to seeparts of Mathcad keeping up with its competition.

Electronic booksWhen Mathcad is installed two electronic books are loadedonto the hard disk, the tutorial and the standard handbook.The handbook contains information drawn from theRubber Handbook'. Although useful, it contains only arelatively small snippet from it. Maybe a future CD-romversion of Mathcad will contain all the non -chemicalinformation from the Rubber Handbook. The tutorial bookon the other hand should prove very useful, especially fornewcomers to Mathcad and those upgrading from the dosversion.

DeficienciesInitially I was very keen about Mathcad but withsubsequent versions my enthusiasm has waned due to itslack of progress. True, the screen presentation is betterunder Windows, and it now runs faster. Computationallyhowever, it cannot do a great deal more than the dosversion 2 release.

The Symbolic Calculator is certainly novel and couldserve as a powerful teaching aid. However it is the absentfeatures that worry me. For instance you cannot directlysolve differential equations - even linear ones. In fact theyare not even mentioned in the user guide. As anypractising engineer knows virtually all modelling isperformed with differential equations.

With Mathcad, if you want to model a simple LCR


PC ENGINEERING

circuit for instance then you would be struggling. One wayis to generate a pair of difference equations from your LCRequation and solve them using matrix notation, Fig. 6,which is very messy.

My reservations regarding 3D plotting have already beenmentioned. If you have a function f(x,y) you should be ableto plot it directly without first having to go through thetiresome process of converting the function into a matrix. Icannot help thinking that in order to get Mathcad to solvea difficult problem you have to spend so much time tryingto evade the limitations imposed by the software.

Operation of the package for complex tasks is notintuitive and becomes rather quirky. For example if youwant to perform an autocorrelation function and you enterthe equation,

rit =Nv Ld-1 A -111-4-Ati

Mathcad will refuse to perform the operation. Everyoneknows that if you have N samples then you are going torun into problems when n+k exceeds N. So why cannot thesoftware issue a warning and perform the calculation up ton+k=N automatically? Instead it stops dead and doesnothing. In order to carry out the task, you have to enterconditionals.

This is typical of problems with the first release fiveyears ago. One does not expect to have the samedeficiency five years on. There is also a concern regardingmemory consumption. It is questionable whether Mathcadmakes as efficient use of memory as it should. Forexample when generating data for a graph, should it usehigh numerical precision for something which is purelyvisual?

User guideThe user guide is a single volume manual which isindispensable - even for the veteran user of the product.After several iterations it has become a well crafteddocument whose layout is appealing and opulentlyillustrated with valuable screen dumps.

As stated, much of Mathcad's operation is rather quirkyand the guide is invaluable for learning the product'speculiarities. Maybe Mathsoft should issue the user guideas a text book as I believe there is a ready market for it.

ConclusionWhen Mathcad was first introduced about five years ago itwas novel, and in its day powerful. It was very popularbecause of its ease of use and its graphic front end.

Several versions later its presentation has improved andits compatibility with Windows makes it an attractivepackage. But it is still fundamentally a maths tool forsolving small scale linear problems only. It has not grownwith expectation.

The main redeeming features are the additions ofSmartMath and the electronic books. What Mathcad doesit does very well, but what worries me is what it does notdo. However for small scale linear modelling there is nobetter product on the market and this new version willsurely find many enthusiasts - especially among thestudent population.

ReferenceCRC handbook of chemistry and physics, CRC Press,(regularly updated).

Mathcad - [ARRAYEMCD1File Edit Text Math Graphics Symbolic Nindow Books Help

Selected Text

exio).4)lee

4fIxIMers LIU ELITines N Roman II

c.2 LillAl etjuj14 Lti

14ELEME6

de50 iteteis

f 3 Mks-

cx.105

abloblab

.--number of elexents---intereleeent spacIng.--frequency---progressive phase ahi in degr

Page 1I

Fig. 5. Polar plots are useful for displaying field patterns from radiators.

x

Vr

A

A

Mallhcad - ILCRMCDIFile Edit Text Math Graphics Symbolic Window Books Help

Variables III-41Y, mcn

auto

IHatTines New Roman 10

Modelling a simple LCR Circuitlobaublabf

Increment stepDamping factor

d 2 d--zy +2 t. -y+ 2 y =OR -g Frequencydt dt Sample interval

This equation can be converted into a pair of dilbresceequations which are represented in a nafrix form

';u+ -2 t t t 402 ) (gui

Y,,,+ t Y.

n:=0 030

411,1

Page 1 T

Fig. 6. In Mathcad, modelling an LCR circuit is difficult. One method is togenerate a pair of difference equations and solve them using matrix notation.

SYSTEM REQUIREMENTS

PC compatiole (Apple version also available),386 or 486 processor, maths coprocessor,Microsoft mouse, Windows , 4Mbyte ram,10Mbyte hard disk, 8MEyte virtual hard diskspace.

SUPPLIER DETAILS

At £375 plus £10 delive-y, Mathcad isavailable from Adept Sc entific MicroSystems, 6 Business Centre, West AvenueOne, Letchworth Hertfo-dshire S.36 2HB.Tel. 0462-L80055, fax. 0462-80213. Appleversion is same price.

I

I auto


DESIGN BRIEF

Table 1. Performancesummary of MAR series

amplifiers. (The colour dotreferred to, in addition to

denoting the type number,also indicates the input

lead).

Ready to use14 amplifiersElectronic component design has yet to reach the point where rfbuilding blocks can be applied as easily as their logic counterparts. AsIan Hickman shows, the MAR series of rf amplifiers require theminimum of skill to apply.

While it may pay to design a clever rfamplifier stage, especially ifconstraints such as a currentconsumption apply, an off-the-peg

solution can be attractive. In these circumstances, theready -to -use rf amplifiers described here can fill thebill perfectly.

The devices referred to are the MAR series fromMini Circuits (a Division of Scientific ComponentsCorporation) and the performance offered by thevarious members of the family is summarised inTable 1. It is an open secret that these are basically anAvantec range of components, but sold by MiniCircuits at supermarket prices, making them anattractive buy.

One way to take a quick look-see what an amplifiercan do is to connect its output back to its input, toimplement an oscillator. As Fig. 1 shows, theintegrated two -stage amplifier is inverting, thecomponent values having been carefully designed togive a nominal match to 5052 at both input and output(type MAR8 excepted). Thus it will oscillate at afrequency Fo when its output is connected back to itsinput via a length of 50Q coax whose electrical lengthis X/2 at F0, as shown in Fig. 2a. With the length ofcoax shown and assuming it has a wave velocity of

65% that of free space, the expected frequency is 0.65x 300/(2 x 0.965) = 101MHz. As the loss in thefeedback network is negligible, the excess loop gain isvirtually equal to to stage's forward gain, so thewaveform would not be expected to be very good, asFig. 2b confirms. Given that the Ins wide spikes onthe edges of the waveform are way beyond the250MHz bandwidth of the oscilloscope, their trueamplitude must be even more horrendous than itappears.

The hard limiting in the amplifier is also responsiblefor excess phase delay (at this frequency the deviceexhibits 9° of excess phase anyway, even under smallsignal conditions) or, put another way, the circuit isalmost a relaxation oscillator, which always results ina lower frequency of oscillation than if the loop gainbarely exceeds unity. In consequence, the actualfrequency of oscillation is less than 100MHz, Fig. 2c,which shows high amplitudes of harmonics: the tenthharmonic is as large as the second, both barely morethan 20dB down on the fundamental, while the thirdharmonic is only 8dB down - definitely not a cleanoscillator.

Fig. 3a shows the interesting effect of reducing thesupply voltage to the circuit of Fig. 2a from +12 to+8V. The narrow spikes are no longer so evident, but

GAN, dBTypical (at MHz)

MAXIMUMPOWER, dBm

DYNAMICRANGE

hterceptpt. dBm

MAXIMUMRATING

DCPOWERat Pin 3

Model (Note \ Output Input NF 3rdNo. Color FREQ. k 4 ) (1d8) Corn- (no dB Order VSWR (25°C) Current Volt.

Dot MHz 100 500 1000 2000 MIN pression damage) Typ. Typ. h Out I(mA) P(mW) (mA) Typ.

MAR -1 Brown DC -1000 18.5 17.5 15.5 - 110 0 +10 5.0 15 1.5:1 1.5:1 40 103 17 5MAR2 Red DC -2300 13 12 8 12.5 11 8.5 + 3 +15 6.5 18 1.3:1 1.6:1 60 325 25 5MAR.3 Orange DC -2300 13 12.8 12.5 10.5 8.0 A+ 8 +15 6.0 23 1.6:11.6:1 70 400 35 5MAR -4 Yellow DC -'COO 8.2 82 8.0 - 7.0 +11 +15 7.0 27 1.9:1 2.1 85 500 50 5MAR -6 write DC -2300 20 19 16 11 9 0 +15 2.8 15 21 1.8:1 50 203 16 3.5MAR -7 Violet DC -270X0 13.5 13.1 12.5 10.5 8.5 + 4 +15 5.0 20 2:1 1.5:1 60 275 22 aMAR8 Baia DC -1000 33 28 23 - 19 +10 +15 3.5 27 65 500 36 7.5


DESIGN BRIEF

Fig. 1 a) The internal circuit of an MAR seriesamplifier. The resistor Rc is not part of the

device, but provides an external DC feed path,while doubling as an rf choke.

020T y p.

b) Biar resistor values for MAR amplifiers. An rf choke maybe advisable in addition, in those cases where the value of Rcis fairly low (i.e. with the lower supply voltages).

AmplifierBias

CurrentBias

VoltageApproximate BiasResistor (Ohms)

ResistorDissipation

(Watts)

is (mA) +V0 +5V +9V +12V +15V +vcc=12V

MAR -1 17 -5 - 235 412 588 .12MAR -2 25 -5 - 160 280 400 .18MAR -3 35 -5 - 114 200 286 I .25MAR -4 50 -6 - 60 120 180 .30

MAR -6 16 -3.5 98 344 531 719 .14MAR -7 22 -4 45 227 364 500 .18MAR -8 36 -8 - - 111 194 .14

COLORDOT

4

RF Outputand Bias

2

Ground Ground

RF Input

.Min.Mut in.

NOTESNOTES (UNLESS OTHERWISE SPECIFIED)

IN1 DIMENSIONS ARE IN

Liz .0102 TOLERANCES -

x

n

96.5 cm

+V

To spectrum analyser(or 'scope via 20 dB pad)

1 n 4701-9-NWY-0-

Fig. 2 a) Connecting the amplifier's outputback to its input via a halfwavelength ofcoax causes it to oscillate. In view of thewildly excessive loop gain, this circuit issheer cruelty to helpless ICs.

Fig. 3 a) Showing the effect of winding thesupply voltage to the circuit of Fig. 2 adown from +12 to +8V, frequency halvingis evident.

b) The waveform produced by the circuit of a.(10mV/div. vertical, 1Ons/div. horizontal

Lb) As a), in the frequency domain.

c) The output of the oscillator, viewed in thefrequency domain. (Ref. level +10dBm buteffectively higher due to the 47052 resistor)vertical 10dB/div, horizontal 100MHz/div,1MHz IF bandwidth, video filter off

c) With further change of supply volts,behavious is even more chaotic.


DESIGN BRIEF

+12 V

10n 1nTo spectrum analyser(or 'scope via 20 dB pad)

1n+12 V

Fig. 4 a) Circuit of Fig. 2 modified to reduce the excessive loop gain.With R = 1500 the circuit oscillates, but not with R = 22052. Secondharmonic 30dB below fundamental, third harmonic 20dB down. Thereduced output loading due to R enables the amplifier to supply almostits rated output power to an external 5051 load circuit.

c) Performanceof a plus b,

showing theclean output

obtained. Secondand third are the

only significantharmonics, bothwell over 30dB

down. (Ref. levelOdBm, other

settingsunchanged.

560 R

MAR 6

560 R

To S.A. or'scope

TLC 271 .1

HP 5082-2303

8K2

b) An add-on to a, to define the amplitudeof oscillation.

1n+12 V

560 R

I Ic) Output of

85MHz crystalcontrolled

oscillator. (Reflevel +10dB,50MHz/div

horizontal, 0Hzat LHS, other

settingsunchanged.

1nTo spectrum analyser(or scope via 20 dB pad)

Fig. 5 a) This circuit, withfrequency determined by atuned circuit, should offerimproved spectral purity, 51 R 150 R

i.e. reduced close -in noisesidebands.

56 RVW*

560 A

MAR 6

XTAL

+12V

Adjustablereference

supply

d) Performanceof a alone withreduced linelength, givingan operatingfrequency of930MHz. (Reflevel OdBm,centrefrequency930MHz,100MHz/div,horizontal,other settingsunchanged.

To spectrum analyser(or 'scope via 20 dB pad)

1n

20 pF.

b) Circuit modified for2 - 10 pF control by a series

4T resonant overtoneSiemens core crystal.B62152-A7X17AL = 18 nH/Turn2

Includes 'scope probe capacitance

d) Outputwaveform ofthe 85MHzcrystaloscillator.


DESIGN BRIEF

circuit behaviour is beginning to be chaotic. It exhibitsa tendency to frequency halving (clearly shown in thefrequency domain in Fig. 3b - the circuit has reachedthe "first bifurcation point", described in Ref.l. Afurther slight change in supply volts prompts evenmore chaotic behaviour, with wide noise -likesidebands appearing around the fundamental andharmonics and the seventh harmonic actually greaterthan the fundamental, Fig. 3c.

To produce a more sanitary oscillator, attenuationwas added in the feedback loop, to reduce the excessloop gain, Fig. 4a. With R = 220Q, the circuit did notoscillate, but did so with R = 15052. The amplitudecontrol loop of Fig. 4b was grafted on and theperformance was then as illustrated in Fig. 4c. Theoutput is very clean; second and third are the onlysignificant harmonics, both well over 30dB down. Thecircuit of Fig. 4a was then run without the amplitudecontrol loop and with a 2 - 10pF trimmer in parallelwith the 5152 resistor in the feedback network. Thisprovided a 7MHz tuning range, and it was noticeablethat at the 10pF setting, the harmonics weresubstantially lower than at 2pF, due to filtering actionon the feedback signal; the levels of 2nd - 4thharmonics seen were respectively -22dB, -28dB,-48dB, higher harmonics being negligible.

To check the performance obtainable at muchhigher frequencies with such a crude and simpleoscillator, the circuit of Fig. 4a was used, with thetrimmer removed and the length of line drasticallyreduced. The circuit oscillated at 930MHz (Fig. 4d)which shows how well the gain of the MAR6 ismaintained with frequency, since R was still set at15052. Note, however, that the amplitude wassubstantially reduced. The frequency of oscillationwas well below that predicted by the line length, dueto the excess phase shift through the amplifier at thisfrequency, amounting to some 70° according to thedata sheet.

Crystal controlThe spectral purity of the simple oscillators describedabove, with their frequency controlled by a length oftransmission line, will not of course compare with thatobtainable with an oscillator controlled by a high Qtuned circuit. This is because in the latter case, thechange of phase shift around the loop with change offrequency is much more rapid than with a 7,/2transmission line. Fig. 5a shows a possibleconfiguration with the necessary 180° phase reversalprovided by a tuned circuit, provided with matchingfor both ports of the amplifier.

Even greater stability and spectral purity will resultfrom crystal control and an 85MHz crystal wasconnected into the circuit of Fig. 5b. Being anovertone crystal operating at series resonance, itcannot conveniently be arranged to provide a phasereversal in the same way as a parallel resonant crystal,so a small two hole balun core was used to provide thephase reversal. It was also arranged to step up theimpedance presented by the crystal circuit to theamplifier's output, while a series tuned circuit set toresonate at the desired frequency was interposedbetween the amplifier output and the reversingtransformer, to suppress oscillations at any but theintended overtone. Excess loop gain was avoided byfitting a pad between the other side of the crystal andthe amplifier's input. As Fig. 5c shows, the circuitproduced an output of -7dBm with low harmoniccontent, the waveform being shown in Fig. 5d. This is

C6

Rin .50,4

01BYPASS

CAPACITOR

01 MAR -3

SUPPLY

< EXTERNAL< BIAS RESISTOR

RF CHOKE

R out = 50,4

PROVIDES BIASFOR TOP PAIR

01 F

Fig. 6 a) Beingunconditionallystable, MARseries amplifiersmay be simplyparalled forhigher outputpower.

b) A parallelpush-pullarrangementoffers lowereven orderharmonicoutput.

c) A push-pullstage may beunilateralisedfor greatlyincreasedreverseisolation.

visibly a bit "secondish", not so very different from anasymmetrical triangular wave. By taking the outputfrom the point shown, it has not had the benefit of thefiltering action of the frequency selective components.An output with lower harmonic content could beobtained from an additional buffer connecteddownstream of the series tuned LC circuit, or evenfrom the pad downstream of the crystal.

Other considerationsThe experiments suggest that these amplifiers aredelightfully tame and easy to apply, provided the twoearth leads are connected directly to a ground plane.Microstrip construction is recommended with alltransmission lines and ground plane running flush tothe package, which means mounting it in a hole in theboard. But for the experiments reported here, a fairlycavalier approach was adopted. The device was sat ontop of the ground plane and the two ground leads werecranked down to connect to connect to it, resulting inlead lengths of a millimeter or more, while othercomponents were mounted using fresh air. Even so,


DESIGN BRIEF

Fig. 7 a) Comparing the performance ofa single ended stage with that of push-pull stages with and withoutunilateralisation.b) Level of second harmonic output,dBc, for the push-pull stage.c) Level of second harmonic output,dBc, for the single ended stage.

-10

-90

Configuration Freq.(MHz)

Gain(dB)

P_ 1dB

(dBm)2nd Harmonic @ P-1dB

(dB below carrier)

Single -ended 100 12 +10 -15Push-pull 100 15 +13.5 -26(Unilateralized)

Push-pull 100 12 +17 -34

100 200 300 400FREQ (MHz)

Push-pull

-10

-90

no problems of instability were encountered.The results indicate that when the amplifiers are run

at well below their output compression point, secondharmonic distortion predominates (e.g. Fig. 5c). Withoverdrive resulting in heavy compression, thirdharmonic is the largest, as the waveform approaches asquarewave, Fig. 2b and c for instance. Where it isdesired to obtain more output than available from asingle device while retaining low harmonic levels,MAR series amplifiers may be paralleled as in Fig. 6a- this is possible since they are (MARS excepted)unconditionally stable. The input and outputimpedances of the paralleled amplifiers fall within therange that is conveniently accommodated by standard4:1, 9:1 and 16:1 broadband line transformerconfigurations. The bandwidth of such a paralleledstage will be limited by the bandwidth of thenecessary matching transformers. In narrowerbandwidth applications, other matching andcombining techniques, such as quarterwavetransformers and n -way Wilkinson splitters/combinerscan be considered.

Note that the gain of such a compound amplifier isonly the same as that of the component individualamplifiers, so to get the desired increased output,additional drive power must be applied. Paralleledamplifiers offer only increased output power, notlower levels of harmonics (unless derated). The fouramplifiers in Fig. 6a could advantageously beredeployed into the slightly more complex circuit ofFig. 6b. In this circuit, owing to the push-pullarrangement, even order harmonics will tend to cancelout.

A push-pull pair of MAR series amplifiers also lendsitself to unilateralisation, Fig. 6c. (Unlikeneutralisation, in which only the reactive componentsof the devices' internal feedback are cancelled,

100 200 300FREQ. (MHz)Single -ended

400

unilateralisation is a technique in which both the realand the imaginary terms of the feedback elements arecancelled. Consequently, unilateralisation tends to beeffective over a wider frequency range thanneutralisation.

The reason that these amplifiers are so easily andeffectively unilateralised is that the Q of their internalfeedback paths is low compared to conventionalamplifiers. In an amplifier that has been unilateralised,the reverse isolation is greatly increased, so thatvariations in the load impedance will no longer affectthe input impedance nor variations in the sourceimpedance affect the output impedance. Butunilateralisation tends to increase both the amplifier'sinput and output impedances, so careful attentionmust be paid to the effects of unilateralisation on inputand output match.

Fig. 7a compares the performance of a single endedMAR series amplifier with that of a push-pull pairwith and without unilateralisaton. It is clear that theonly major advantage of unilateralisation is theincreased reverse isolation, the straighforward push-pull pair being better on other counts. In particular,unilateralisation has lowered the 1dB compressionpoint by 3.5dB. This is partly due to the power lost inthe resistive components of the cross -coupledfeedback networks and partly to the effect on inputand output impedances. Fig. 7b and c compare thelevel of second harmonic in dBc for the single endedand push-pull amplifiers respectively.

References.1. Chaos and engineering R. Dettmer IEE Review Sept.1993 pp 199 - 203Mini Circuits. For further information contact Tel. 0252835094 .


M & B RADIO (LEEDS)THE NORTH'S LEADING USED TEST/EQUIPMENT DEALER

OSCILLOSCOPESTEKTRONIX 2445A 150MHZ GPIB T1500TEKTRONIX 475 200MHZ DUAL TRACE WITH PROBES 000TEKTRONIX 466 I 00MHZ STORAGE . L395TEKTRONIX 5C504/761503/0M501 PORTABLE 80MHZ SCOPE

L650KIKUSUI C06100 100MHZ 5 CHANNEL L5001WATSUI SS6122 100MHZ 4 TRACE CURSOR READOUT..........L900HITACHI V222 20MHZ DUAL TRACE WITH MANUAL L225GOULD 054000 10MHZ DIGITAL STORAGE ... 1I95

FARNELL DTC12 I 2MHZ DUAL TRACE/COMPONENT TESTER L175FARNELL DT12/5 12MHZ DUAL TRACE . . £150HP 1722B 275MHZ MICROPROCESSOR CAL MEASUREMENTS . . L750HP I725A 275MHZ OSCILLOSCOPES (AS NEW) L600TEKTRONIX 46SA/B 100MHZ DUAL TRACE PRICES FROM L300

BECKMAN 9020 20MHZ DUAL TRACE . . L225GOULD 05300 20MHZ DUAL TRACE.. ... LI95TELEQU1PMENT D34 I 5MHZ PORTABLE . LI50NICOLET 4094A DIGITAL OSCILLOSCOPE £500

SIGNAL GENERATORSMARCONI 2019 80KHZ TO 1040MHZ . . . L1900MARCONI 20I8A 80KHZ TO 520MHZ . L1250MARCONI 2018 80KHZ TO 520MHZ SYNTHESIZED L750MARCONI 2008 10KHZ TO S20MHZ INC RF PROBE KIT

MARCONI 2015/2171 SYNCHRONIZER IOMHZ TO 520MHZ (ASNEW) £325

HP 864015 500KHZ TO 550MHZ OPT 001 ... 050HP 8616A I 8GHZ TO 4 5GHZ £295

HP 4204A 10KHZ TO I OMHZ L250HP 8165A .001 HZ TO 50MHZ PROD' SIGNAL GENERATOR............. L950ADRET 740A 10KHZ TO I.IGHZ SYNTHESIZED 0500FARNELL SSG2000 10KHZ TO 2GHZ SYNTHESIZED (AS NEW) £2250FARNELL SSGI000 OKHZ TO I GHZ SYNTHESIZED (AS NEW) LI500FARNELL SSG520 IOMHZ TO S20MHZ 8 TTS 520 TRANSMISSION

ADRET 20230A I MHZ SYNTHESIZED SOURCE ............. 6195POLRAD 1106ET 1.8 TO 4.6GHZ WITH MODULATOR L400RHODE A SCHWARZ TYPE SMC I 4.8GHZ TO 1 2.6GHZ L500RHODE & SCHWARZ NOISE GENERATOR I TO 1000MHZ SKTU

(NEW) L150WAVETEK 193 20MHZ SWEEP FUNCTION GENERATOR. L295MARCONI 2020 50KHZ TO 520MHZ SYNTHESIZED L500MARCONI 2000 20KHZ TO 20KHZ WITH METERED

ATTENUATOR L125TEXSCANSAYROSA FREQUENCY OS

WEEPGECILLATOR MA30 10HZ TO

10KHZ... £250EXACT 628 20MHZ PULSE/SWEEP FUNCTION

GENERATOR £650G1GA GRI IOTA 12 TO 18GHZ PULSE GENERATOR (NEW) £750

TEST EQUIPMENTEFRATOM FRT ATOMIC FREQUENCY STANDARD. L2500HP 5065A RUBIDIUM VAPOR FREQUENCY STANDARD £2000

MARCONI 2955 COMMUNICATIONS TEST SET £3000MARCONI 2305 MODULATION METER 0000MARCONI TF230015 MODULATION METER L395MARCONI 2331 DISTORTION FACTOR METER L250FARNELL LA.520 RE AMPLIFIER I.5MHZ TO 520MHZ L150FARNELL LFM4 AUDIO OSCILLATOR L200FARNELL TP18 TRUE RMS SAMPLING RF METER AS NEW, I GHZ L350TEKTRONIX 52IA PAL VECTORSCOPE .L1000DYMAR 2085 AF POWER METER .. L225BIRD TENULINE 8143 100W 6DB ATTENUATOR (NEW) . . L100NARDA 3022 01 DIRECTIONAL COUPLER 1GHZ TO 4+31-1Z .. L250NARDA 3001-30 DIRECTIONAL COUPLER 460MHZ TO 960MHZ £100AVO RM215 U2 AODC BREAKDOWN TESTER L300ROTEK AC/DC PRECISION CALIBRATOR............... .

EXACT 334 CURRENT PRECISION CALIBRATOR 1I95HP I I 667A POWER. SPLITTER (NEW) 18GHZ ............ L600

L300....L100

L700£700

...000

NARDA 769/6 150VY 6DB HIGH POWER ATT (NEW) _

RACAL DANA 5002 WIDEBAND LEVEL METERRACAL DANA 9301 TRUE RMS RF LEVEL METER ......,..RACAL DANA 6000 MICROPROCESSING DVM .

RACAL DANA 9000 MICROPROCESSING TIMER COUNTERL275

RACAL DANA 488 IEEE STD 488 BUS ANALYSER L250RACAL DANA 1002 THERMAL PRINTER. 050RACAL DANA 9302 RE MILLIVOLTMETER 10KHZ TO I 500MHZ L400RACAL 930IA RF MILLIVOLTMETER 10KHZ TO 1500MHZ. £250

RACAL DANA 1998 1 3GHZ FREQ COUNTER/TIMER L750RACAL 9841 3GHZ FREQUENCY COUNTER. 025RACAL 9921 3GHZ FREQUENCY COUNTER. ..... L400RACAL 9903 TIMER COUNTER L100

RACAL 9915 10HZ TO 520MHZ FREQUENCY COUNTER ............... CI50RACAL 9919 1.109Z FREQUENCY COUNTER/TIMER L325HP 43SA POWER METER 8482H HEAD L650BICC/TEST T431M CABLE TEST SET L250

FARNELL B30/5 POWER SUPPLY 030VOLT SAMP...._ ..645FARNELL 830/20 POWER SUPPLY 0-30VOLT 20AMP. ........ 6100RACAL 9063 TWO TONE GENERATOR.....L250FARNELL ELECTRONIC LOAD RB 1030/35 L495FARNELL TRIPLE OUTPUT POWER SUPPLY TOPS 3D . £225

MARCONI TF2910/4 TV LINEAR DISTORTION ANALYSER . L750MARCONI TF291 I TEST LINE GENERATOR + INSERTER L750MARCONI TF29 IRA INSERTION SIGNAL ANALYSER £750

FEEDBACK ssoeor SINE/SQUARE AUDIO OSCILLATOR...................085FEEDBACK 602 YARIABLE PHASE OSCILLATOR _ L145FEEDBACK SEGO] 1 I 2MHZ SWEEP FUNCTION GENERATOR £245FEEDBACK TFA TRANSFER FUNCTION ANALYSER. . . . . L395FEEDBACK EW604 ELECTRONIC WATTMETER L195HP 3455A HIGH STABILITY VOLTMETER GPIB L1200HP 8405A VECTOR VOLTMETER & ACCESSORIES.. £150HP 8406A COMB FREQUENCY GENERATOR L250HP 3468D DIGITA_ MULTIMETER . . L400HP 3466A DIGITA- MULTIMETER L200HP 8750A STORAGE NORMALIZER £395HP 3400A TRUE RMS VOLTMETER. C145

HP 3403C TRUE RMS VOLTMETER DIGITAL..........._ . LISOHP 394A ,ARIABLE ATTENUATOR I GHZ TO 2GHZ L85HP 3406P SAMPLING VOLTMETERHALCYCN 5006/521A UNIVERSAL TEST SYSTEM ............ ............. L650TEKTRONIX 576 CURVE TRACER & 172 PROGRAMMEABLE TEST

PSI A100 WAVEFORM GENERATOR .... L225

1WATSUI IGHZ FREQUENCY COUNTERHP 5006A, SIGNATURE ANALYSER..........KEMO D41 PHASE METER I HZ TO 100KHZ _

WAYNE KERR CT496 LCR BRIDGE PORTABLEMARCONI TF2700 LCR BRIDGE PRICES FROMPHILIPS.M8252A DUAL PEN RECORDER ............HP 324ADISTORTION METER.FLUKE 73013 PROG CONSTANT CURRENT/VOLTAGE

UNA02.14 FIELD STRENGTH METER ....

1095015

L250

... £750

L200BIRD 43 THROUGHLINE WATTMETERS_ .... _ £145PACIFIC MEASUREMENTS 1034A RF POWER METER I MHZ TO

!BCH:. 0450PHIUPS PM5565 VIDEO WAVEFORM MONITOR L225BALLAMTINE 6125C PROG TIME/AMPLITUDE TESTSET 0500FERROC RAPH RTS2/ATU I AUDIO TESTSETTEKTRONIX 1421 VECTOR SCOPE ..._ _ C100TEKTRONIX 528A VIDEO WAVEFORM MONITOR £300

SIEMENS U2233 PSOPHOMETER (NEW) 0500

SPECTRUM ANALYSERS

HP 8558B 10MHZ TO I.5GHZ WITH 182T MAINFRAME............ L11100HP 1417 85528 IF 135548/8553B RE UNITS 0000HP 3580A 5KHZ TO 50KHZ ANALYSER (NEW) .61750B&K 2013 AUDIO ANALYSER 1HZ TO 20KHZ....MARCONI 2370 1 10MHZ ANALYSER.. C950HP 1417 8553B/8552B 1 IOMHZ WITH 8443A TRACKING GENERATOR

HP 8449A TRACKING GENERATOR 0.5 TO 1300MHZ L1000TEKTRONIX 492 50KHZ TO 18GHZ PORTABLE ..........._._............£5500TEKTRONIX 492AP 10KHZ TO 22GHZ (325GHZ)HP 35111A WAVE ANALYSER I SHZ TO 50KHZ (AS NEW) £850HP 8505A NETWORK ANALYSER 500KHZ TO 1300MHZ ........ ... LS000

RADIO EQUIPMENTRACAL RAI 792 HF RECEIVERS

PHIUPSEIBMHZ TO I (*MHZ BROADCAST TX I OKW........ 63000EDDYSTONE 990S UHF 250 TO 850MHZ L159EDDYSTONE 1990/3 RECEIVER 25 TO 500MHZ.. ............. .... L450ROCKWELL COLLINS HF80 IRA EXCITER (NEW).................. 6900RACAL SYNCAL 30 TRANSCEIVERS. 1300RACAL RA 1 779/MA 1 107 HF RECEIVERREDIFON 8500 SYNTHESISED HE RECEIVER LSOO

BULK l'URCHASE SPECIALSMARCONI 1101 RC OSCILLATOR 20HZ TO 20KHZ L45ADCOLA 777 DESOLDERING STATION . C100

ALL PRICES PLUS VAT AND CARRIAGE86 Bishopsgate Street, Leeds LS I 4BB

Tel: (0532) 435649 Fax: (0532) 426881CIRCLE NO. 110 ON REPLY CARD

Custom metalworkgood & quick!

Plus a wide range of rack -mounting

Send for our new Catalogue.cases etc. from stock.

-,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

I rrir BRVIONal SYSIIMS

IPK Broadcast Systems3 Darwin Close Reading Berks RG2 OTBTel: (0734) 311030 Fax: (0734) 313836

THIS IS THECOUPONthat brings the 1993-94

CATALOGUEI140 pages, A4, copiously illustrated, bangup to date and with bonus vouchers. Sendcheque/PO for £1.50

I that brings the

I SERVICECATALOGUE

14-12Iwith choice of very wide ranges of famous -name qualityelectronic components and associated gear sent promptly onreceipt of your order. A service

I that

Ihave been providing constructors with continuously since 1965.With this Catalogue, you will find our service the best ever. Sendfor your copy NOW!11-

ELECTROVALUE LTD., 3 CENTRAL TRADING ESTATE, STAINES TW18 4171Telephone 0784 442253 Fax 0784 460320

I I enclose Cheque/PO Order/Credit Card NoI value f1.50 for your 1993-94 CATALOGUEI Name

I Address I

I I

E=IWino(alue

Post CodewwJ

CIRCLE NO. 111 ON REPLY CARD CIRCLE NO. 112 ON REPLY CARD


OSCILLOSCOPESHEWLETT PACKARD 54L00A - 100 MHz DIGITISING 11100HEWLETT PACKARD 1140A. 1741A, 1744A - 100Mliz DUAL

CHANNEL 050HEWLETT PACKARD 1707A, 17078 - 75MHz DUAL

HEWLETT PACKARD I82c - 100MHz FOUR

TEKTRONIX 453, 454, 465, 465B, 475, 475A, 466, 468 - 100 to500MHz DUAL CHANNEL from £250

TEKTRONIX 7603, 7613, 7623, 7633, 7313, 7704A. 7844, 7504 -100 to 500 MHz TWO or FOUR CHANNEL.... ... from L350

(Various plug in options available)TELEQUIPMENT DM63, D75, D83-50MHz DUAL

CHANNEL from 1175GOULD MODELS OS 245, OS 4000, OS 4020, OS

4200 from £125GOULD 400-20MHz BAND WIDTH - 100MsS SAMPLING

RATE - AS NEW £1350PHILIPS 3211, 3212, 3217, 3226 3240, 3243, 3244, 3261. 3262 -

15 to 120Mt TWO AND FOUR CHANNELfrom 1125 to 6350

SOLARTRON SCHLUMBERGER CD 1740 - 20Mliz - FOURCHANNEL from £250

DUMART 10r., 50MHz DUAL CHANNEL 6250

SPECTRUM ANALYSERSHEWLETT PACKARD 3580A - 51-1. to 50KHz ....... ..£1250HEWLETT PACKARD 3582/1 - 25KHZ DUALCHANNEL....£3000HEWLETT PACKARD 3585A - 20Hz to 40KHz (GPM)._ L1250HEWLETT PACKARD 182T- MAINFRAME WITH 8559A

I 05111.41z to 21Gliz) . M750HEWLETT PACKARD 8590A - IONIKM to 150 CH.

£4250AILTECH 727 - 20GHz L2200MARCONI 2370 -110MHs /1250ADVANTEST TR4131 - 10KHZ - 3.5 GK. £4500HEWLETT PACKARD 14IT WITH 8552B 4- 8554B (500KHz -

HEWLETT PACKARD 8757C - SCALAR NETWORK (200Hz)ANALYSER L4500

TEKTRONIX 1603 WITH 7L 14 ( 10KHz I SCHz) 12000

MISCELLANEOUSHEWLETT PACKARD 370IA 3102A 3703A MICROWAVE

LINK ANALYSER £1000HEWLETT PACKARD 3702133705A 3710A/37I6A -

MICROWAVE LINK ANALYSER £1500HEWLETT PACKARD 3779A PRIMARY MULTIPLEX

ANALYSER £1000HEWLETT PACKARD 415E - SMIt. METER L275HEWLETT PACKARD 8750A- STORAGE

HEWLETT PACKARD 5342A FREQUENCY COUNTER(113GH.) t1400

HEWLETT PACKARD 8493A TRACKING GENERATOR/COUNTER L450

HEWLETT PACKARD 8445E AUTOMATIC PRE -

HEWLETT PACKARD 4953A - PROTOCOL

HEWLETT PACKARD 3456A DIGITAL VOLTMETER.... £675HEWLETT PACKARD 8011A PULSE GENERATOR........ L500HEWLETT PACKARD 8013B PULSE GENERATOR ......... £750KEITHLY 197AUTO M.M. WITH IEEE £400MARCONI 2830 PCM MULTIPLEX TESTER £1250MARCOM 2831 CHANNEL ACCESS SWITCH E500RACAL -DANA 9081 SYNTHESISED SIGNAL GENERATOR

(520 MHz) L550RACAL -DANA 9084 SYNTHESISED SIGNAL

GENERATOR 6500RACAL -DANA 9303 TRUE RMS - R/F LEVEL METER ._.£650ANRITSU - ME588C - MICROWAVE SYSTEM ANALYSER

(AS NEW).. £3500AVO - RM2 15112 -BREAKDOWN/LEAKAGE 0 IONISATIONTESTER.....£500DATRON 106IA - AUTOCAL MULTI FUNCTIONAL

VOLTMETER (6.5 DIGITS) . . £950DATRON 107IA AUTOCAL MULTI -FUNCTIONAL

VOLTMETER (7.5 DIGITS) ...E1155FARNELL RB1030-35 ELECTRONIC LOAD. . 6450GALLENKEMP 3000 SERIFS THERMOSTAT/CYCLE

TIMERITMPERATURE CYCLING OVEN._.....£750ROHDE SCHWARTZ QDM BN 36711 DIGITAL '0'

METER .E400PHILIPS 5167 1-10 MHz FUNCTION GENERATOR £400PHILIPS PM6672 TIMER COUNTER IGHz WITH IEEE £650SOLARTRON/SCHLUMBERGER 1170 - FREQUENCY

SOLARTRON /SCHLUMBERGER 3530 - ORION DATALOGGING SYSTEM L750

WAYNE KERR 8905- AUTOMATIC PRECISIONBRIDGE 6800

WAYNE KERR 3245 - PRECISION INDUCTANCEANALYSER 62500

WILTRON 560 SCALAR NETWORK ANALYSER... ... T900YOKOGAWA YEW 3655 ANALYSING RECORDER....... L650B K 2511 VIBRATION METER 11250B K - 2512 VIBRATION METER (2511 1621). £2000B & K- 2515 VIBRATION METER (FFT) £4500HEWLETT PACKARD 5335A - UNIVERSAL COUNTER

WITH IEEE £450PHILIPS PM2625 MULTIMETER WITH IEEE £850PHILIPS PM8272 XYT CHART RECORDER £600TEK TM50.3/SG533/P11506/TG.50 SCOPE CALIBRATOR E2200

ANRITSU ML93B OPTICAL POWER METER WITHSENSOR

MANY MORE ITEMS AVAILABLE - SEND LARGE S.A.E. FOR LIST OF EQUIPMENT

ALL EQUIPMENT LISTED HERE IS USED, WITH 30 DAYS GUARANTEE

Please check availability before ordering. Carriage + VAT to be added to all goods

TELNET8 Cavans Way, Binley Ind. Estate, Coventry CV3 2SF

(Premises situated close to Eastern Bypass in Coventry with easy access to Ml, M6, M40, M42, M69, M45)

TELEPHONE 0203 650702 FAX 0203 650773CIRCLE NO. 115 ON REPLY CARD

Z8 Z80 Super86502 6581668HC1168000...

from only £698051...and many moreIC CD _

C_MOBB=

LA7 CD

J LS

23LE

Call for data sheetsTel: 081-441 3890Fax: 081-441 1843 S MARTCOMMUNICATIONS


Low cost data acquisition for IBM PCs & compatiblesAll our products aro easy to inshdi - theyeither the printer or serial port end rThey ere supplied wilts easy to was sofor either display or print -eat.

ADC -10

ADC -1 1

ADC -16

8 - bit resolution

one channel

10-25K samples per second

Oscilloscope/Voltmeter software

0-5V input range

Connects to printer portf49 10 - bit resolution

11 channel

5-10K samples per second

Data logger software

0-2.5V input range

Connects to printer port

£75 8, 12, la - bit resolution - sign 8 s/e or 4 differential inputs

216 or 3008 -bit samples per second

2.5V input range

Data logger software

Connects to serial port

All wives *Madre

f99PICO TECHNOLOGY LTDBroadway Noun, 149 151 SI Hoots Road, Nardwkk, Cambridge CH 70.1

VISA TEL: 0954 211716 FAX: 0954 211880NIMPSI

KESTRELELECTRONIC

COMPONENTS LTD* All items guaranteed to manufacturers' spec.* Many other items available.

'Exclusive of V.A.T. and post and package'

1+ 100+ 1+ 100+

EPROMS STATIC RAMS2764A-250 2.50 2.20 62256ALP-10 3.30 2.6027C64-150 2.50 2.20 6264ALP-10 1.95 1.4827128A-200 2.40 2.20 6116ALP-10 1.10 0.8027256-250 2.30 2.10 628128LP-10 12.00 10.0027C256-150 2.90 2.30 HD63B21P 1.95 1.5027C512-150 3.30 2.55 6522P 2.40 1.8027C010-150 5.80 4.20 65CO2P2 2.90 2.5027CO20-150 9.00 7.60 65C21P2 2.90 2.5068B50P 1.40 0.85 65C22P2 2.75 2.40D8749H 4.40 3.75 Z80A CPU 1.30 0.99MM58274CN 7 00 3.95 Z80A P10 0.95 0.7580C31-12meg 2.60 2.10 Z80A CTC 0.90 0.7075176BP 1 90 0.95 Z80A DART 2.20 1.38

74LS, 74HC, 74HCT Series availablePhone for full price list

All memory prices are fluctuating daily, please phone toconfirm prices

178 Brighton Road,Purley, Surrey CR8 4HA

Tel: 081-668 7522. Fax: 081-668 4190.CIRCLE NO. 117 ON REPLY CARD

136



In the penultimate part of his series on poweramplifier design, Douglas Self presents a workedexample delivering 0.0015% total harmonicdistortion. But this performance is only attainablewith correctly chosen compensation... -

COisticovriticxriin power amplifiers7: Frequency compensationand real designs

The distortion performance of an ampli-fier is determined not only by open looplinearity, but also the negative feedback

factor applied when the loop is closed. In mostpractical circumstances doubling the NFB fac-tor halves the distortion. To date, this serieshas focused on basic circuit linearity. I haveassumed that open loop gain falls at6dB/octave due to a single dominant pole,with the amount of NFB permissible at hfbeing set by the demands of hf stability.Because of this, the distortion residuals from a`blameless' amplifier are comprised almostentirely of crossover artifacts due to their highfrequency content. Audio amplifiers usingmore advanced compensation are rather rare.However, certain techniques do exist...

This series has stuck close to conventionaltopologies, because even commonplace cir-cuitry has been shown to have little known

FIG 1A MILLER-CAPACITMWITHAM -FELECCPFtNEATION.

aspects and interesting possibilities. Thisimplies a two -gain -stage circuit (unity gainoutput stages not being counted) with most ofthe feedback applied globally, but smoothlytransferred to the voltage amplifier stage aloneas frequency increases. Other configurationsare possible; a one stage amplifier is anintriguing possibility - they are common incmos op -amps - but is probably ill-suited topower amp impedances. See reference 1 for aneccentric three -stage amplifier with an openloop gaM of just 52dB (due to the dogged useof local feedback) and only 20dB of globalfeedback. Most of the section below refersonly to the conventional two -stage structure.

Making a pole dominantDominant pole compensation is the simplestkind, though its implementation involves sub-tlety. Simply take the lowest pole to hand

FIG 1B C131.9.5RTIEN.

(P1), and make it dominant, ie so much lowerin frequency than the next pole P2 that thetotal loop gain (the open loop gain as reducedby the attenuation in the feedback network)falls below unity before enough phase shiftaccumulates to cause hf oscillation. With asingle pole, the gain must fall at 6dB/octave,corresponding to a constant 90° phase shift.Thus the phase margin will be 90° givinggood stability. Figure 1a shows the tradition-al Miller method of making a dominant pole.The collector pole of Tray is lowered by addingthe Miller capacitance Cdrn to that which

Fig. 1. Implementing dominant -polecompensation. (a) Miller capacitor, (b) Shunt -lagcircuit (c) Output -stage Inclusive Millercompensation. (d) How to implement 2 -polecompensation. Seep 140.

LUIRLIT-STAGE-IPCLIERVE

FIG MILLER CEMPEASATI114.

FIG 1D COM2-PCLE.DEATICN.

February 1994 ELECTRONICS WORLD + WIRELESS WORLD I 37

AUDIO

Fig. 2. The principle of nested feedback loops.

unavoidably exists as the Cbc of the VAS tran-sistor. However there are other beneficialeffects; Cdom causes "pole splitting", in whichthe pole at Tr2 collector is pushed up in fre-quency as P1 moves down - most desirablefor stability. Simultaneously the local NFBthrough Cdom linearises the vas.

Assuming that input stage transconductanceis set to a plausible 5mA/V, and stability con-siderations set the maximal 20kHz open loopgain to 50dB, then from the equations in Part1, Cdom must be 125pF. This is more thanenough to swamp the internal capacitances ofthe vas transistor, and is a realistic value.

The peak current that flows in and out ofthis capacitor for an output of 20V rms,20kHz, is 447µA. Recalling that the inputstage must sink Cdom current while the vascollector load sources it, and likewise the inputstage must source it while the vas sinks it,there are four possible places in which slewrate might be limited by inadequate currentcapacity. If the input stage is properlydesigned then the usual limiting factor is vascurrent sourcing. In this example a peak cur-rent of less than 0.5mA should be easy to dealwith, and the maximum frequency forunslewed output will be comfortably above20kHz.

Figure 1 b shows a much less satisfactorymethod - the addition of capacitance toground from the vas collector. This is usually

called shunt lag compensation, and as PeterBaxandall aptly put it, "The technique is in allrespects sub-optimal2."

We have already seen in Part 3 that loadingthe vas collector resistively to ground is a verypoor option for reducing LF open loop gain,and a similar argument shows that capacitiveloading to ground for compensation purposesis an even worse idea. To reduce open loopgain at 20kHz to 50dB as before, the shuntcapacitor Ciag must be 43.6nF, which is awhole different order of things from 125pF.The current flowing in Ciag at 20V rms,20kHz, is 155mA peak, which is going torequire some serious electronics to provide it.This important result can be derived by simplecalculation, and I have confirmed it with Spicesimulation. The input stage no longer con-strains the slew rate limits, which now dependentirely on the vas.

A vas working under these conditions isalmost certain to have poor linearity. The cur-rent variations in the stage, caused by the extraloading, produces more distortion and there isnow no local NFB through a Miller capacitorto correct it. To make matters worse, the dom-inant pole P1 will probably need to be set to alower frequency than for the Miller case, tomaintain the same stability margins, as there isnow no pole splitting to raise the pole at theinput stage collector. Hence Ciag may have tobe even larger, and require even higher peak

2POLE2.CIR Conceptual transconductance+current-amp-VAS. 18 Nov 93Date/Time run 11/18/93 23:24:33 Temperature. 25 0

120 T 1

100

80

60

40

20

0+10 100

2-P0 LIB() (R1181)1.

1k 10k

Frequency (Hz)

100kCp1=100pFCp2=220pF

13,3=2k2

1M 10M

Fig. 3.13Spice plot of theopen -loop gainof a 2 -polecompensatedamplifier. Thedifferencebetween thetwo plots showsthe amount ofextra NFBpossible.

currents. Takahashi has produced a fascinatingpaper on this approach3, showing one way ofheaving about the enormous compensationcurrents required for good slew rates. The onlything missing is an explanation of why shuntcompensation was chosen in the first place.

Including the output stageMiller capacitor compensation elegantlysolves several problems at once, and the deci-sion to use it is not hard. However the ques-tion of whether to include the output stage inthe Miller feedback loop is less easy. Suchinclusion (see Fig. 1c) presents the desirablepossibility that local feedback could lineariseboth the vas and the output stage, with just theinput stage left out in the cold as frequencyrises and global NFB falls. This idea is mostattractive as it would greatly increase the feed-back available to linearise a Class B outputstage.

There is certainly some truth in this whereapplying Cdom around the output as well as theVas reduced the peak IkHz THD from 0.05%to 0.02%4. However it should be pointed outthat the output stage was deliberately underbiased to induce crossover spikes because,with optimal bias, the improvement was toosmall to be either convincing or worthwhile.Also, this demonstration used a model ampli-fier with TO -92 "output" transistors. In myexperience this technique just does not workwith real power bipolars because it inducesintractable HF oscillation.

The use of local NFB to linearise the vasdemands a tight loop with minimal extra phaseshift beyond that inherent in the Cdom domi-nant pole. It is permissible to insert a cascodeor a small signal emitter follower into thislocal loop, but a sluggish output stage seemsto be pushing the phase margin too far; theoutput stage poles are now included in theloop, which loses its dependable HF stability.Bob Widlar has stated that output stagebehaviour must be well controlled up to100MHz for the technique to be reliable5. Thiswould appear to be virtually impossible fordiscrete power stages with varying loads.

While I have so far not found "InclusiveMiller compensation" to be workable myself,others may know different. If anyone can shedfurther light I would be most interested.

Nested feedback loopsNested feedback is a way to apply more NFBaround the output stage without increasing the


AUDIO

1'100W

CS

1OLF

IN1V WS

11/CB

mimSOON,

R2

10OR

TR?

RS

1014

TRIOTRU

R7

GBR

INPUT STAGE

global feedback factor. The output has anextra voltage gain stage bolted on, and a localfeedback loop is closed around these twostages. This NFB around the composite blocreduces output stage distortion and increasesfrequency response, to make it safe to includein the global NFB loop.

Suppose that bloc A1 (Fig. 2a) is a distor-tionless small signal amplifier providing allthe open loop gain and so including the dom-inant pole. A3 is a unity gain output stage withits own main pole at 1MHz and distortion ofI% under given conditions: this 1MHz poleputs a firm limit on the amount of global NFBthat can be safely applied.

Fig 2b shows a nested feedback version; anextra gain bloc A2 has been added, with localfeedback around the output stage. A3 has themodest gain of 20dB so there is a good chanceof stability when this loop is closed to bringthe gain of A3+A2 back to unity. A2 now expe-riences 20dB of NFB, bringing the distortiondown to 0.1%, and raising the main pole to

R13

100R

R14

ISR

TR13

TR12

R9

11(

I

VAS

TFi6

'VVAR12 100R

CS

F R1SBOOR

R24 100R

'V AA,

RI6

OR22

+37V

cg

2201E

L1

10 TURNS

1 IN DIN1

R17 RSBOR22 RSB SON

167

100W

OUT

Tr1.3, 5, 14 MPSA56

Tr4, 10-13 MPSA06

Trb MJE340

Tr, MJ802

Tr8 MJE350

Trl M14502

C10

220LF

T -37V

OUTPUT

Fig. 4. 50W Class B amplifier circuit diagram. Transistor numbers correspond It ith the genericamplifier in Part 1.

BIASGEN3.CIR Class -B transistor bias generators current -compensate RTemperature 25.0

1 R14

Date/Time run: 11/18/93 22:54:12

1 55V

1.50v

1.45V

1 40V+Fig. 5. Spice plot of the voltage -peaking behaviour of 4 OmA 4 5mA 5 O mA 5 5mA 6 O mA 6 5mA 7 OmA 7 5mA 8 OmA

a current -compensated bias generator. . o . v (4)

I IN

052

1652

1852

2752


AUDIO

AUDIO PRECISION CLASSB THD+N(Z) vs FREQ(Hz)

0.2

0.1

0,010

0,001

.0005

10 100 1k 10k 50k

Fig. 6. Class 8 amplifier: THD performance at 50W/8 -ohm; measurement bandwidths 30kHz and80kHz.

AUDIO PRECISION CLASSB THD+N(Z) vs FREQ(Hz)

0.2

0,1 ..., 11

Fig.7. The dramatic THD improvement obtained by converting the Class B amplifier to 2 -pole

10MHz, which should allow the application of20dB more global NFB around the overallloop that includes Al. We have thus decreasedthe distortion that exists before global NFB isapplied, and simultaneously increased theamount of NFB that can be safely used,promising that the final linearity could be verygood indeed. For another theoretical examplesee reference 6.

Real life examples of this technique inpower amps are not easy to find, but a varia-tion is widely used in op -amps. Many of uswere long puzzled by the way that the muchloved 5534 maintained such low THD up tohigh frequencies. Contemplation of its entrailsappears to reveal a three -gain stage designwith an inner Miller loop around the thirdstage, and an outer Miller loop around the sec-ond and third stages; global NFB is thenapplied externally around the whole lot.Nested Miller compensation has reached its

apotheosis in cmos op -amps - the presentrecord appears to be three nested Miller loopsplus the global NFB7. Don't try this one athome.

Two pole compensation

Two pole compensation is a mildly obscuretechnique for squeezing the best performancefrom an op-amp8'9, but it has rarely beenapplied to power amplifiers. I know of onlyone examples. An extra HF time constant isinserted in the Cdom path, giving an open loopgain curve that initially falls at almost12dB/octave, but which gradually reverts to6dB/octave as frequency continues to increase.This reversion is arranged to happen wellbefore the unity loop gain line is reached, andso stability should be the same as for the con-ventional dominant pole scheme, but withincreased negative feedback over part of theoperational frequency range. The faster gain

roll off means that the maximum amount offeedback can be maintained up to a higher fre-quency. There is no measurable mid bandpeak in the closed loop response.

One should be cautious about any circuitarrangement which increases the NFB factor.Power amplifiers face loads that vary widely:it is difficult to be sure that a design willalways be stable under all circumstances. Thismakes designers rather conservative aboutcompensation, and I approached this techniquewith some trepidation. However, results wereexcellent with no obvious reduction in stabil-ity. Figure 7 shows the result of applying thistechnique to the Class B amplifier describedbelow.

The simplest way to implement two polecompensation is shown in Fig 1 d, with typicalvalues. Cpi should have the same value as itwould for stable single pole compensation,and Cp2 should be at least twice as big; Rp isusually in the region 1k -10k. At intermediatefrequencies Cp2 has an impedance comparablewith Rp, and the resulting extra time constantcauses the local feedback around the vas toincrease more rapidly with frequency, reduc-ing the open loop gain at almost 12dB/octave.

At HF the impedance of Rp is high com-pared with Cp2, the gain slope asymptotesback to 6dB/octave, and then operation is thesame as conventional dominant pole, withCdom equal to the series capacitance combi-nation. So long as the slope returns to6dB/octave before the unity loop gain crossingoccurs, there seems no obvious reason whythe Nyquist stability should be impaired.

Fig. 3 shows a simulated open loop gainplot for realistic component values; Cp2 shouldbe at least twice Cpi so the gain falls back tothe 6dB/octave line before the unity loop gainline is crossed. The potential feedback factorhas been increased by more than 20dB from3kHz to 30kHz, a region where THD tends toincrease due to falling NFB. The open loopgain peak at 8kHz looks extremely dubious,but I have so far failed to detect any resultingill effects in the closed loop behaviour.

There is however a snag to the simpleapproach shown here, which reduces the lin-earity improvement. Two -pole compensationmay decrease open loop linearity at the sametime as it raises the feedback factor that strivesto correct it. At HF, Cp2 has low impedanceand allows R to directly load the vas collectorto ground. This worsens vas linearity as wehave seen. However, if Cp2 and Rp are cor-rectly proportioned the overall reduction indistortion is dramatic and extremely valuable.When two pole compensation was added toFig. 4, the crossover glitches on the THDresidual almost disappeared, being partiallyreplaced by low level 2nd harmonic whichalmost certainly results from vas loading. Thepositive slew rate will also be slightly reduced.

This looks like an attractive technique, as itcan be simply applied to an existing design byadding two inexpensive components. If Cp2 ismuch larger than CH, then adding/removingRp allows instant comparison between the twokinds of compensation. Be warned that if an


AUDIO

amplifier is prone to HF parasitics then thiskind of compensation may exacerbate them.

Design example: a 50W class B amplifierFigure 4 shows a design example of a Class Bamplifier intended for domestic audio. Despiteits conventional appearance, the circuit deliv-ers a far better distortion performance thanthat normally associated with the arrangement.

With the supply voltages and values shownit delivers 50W/8Q from IV rms input. In pre-vious articles I have used the word blamelessto describe amplifiers in which all distortionmechanisms, except the apparently unavoid-able ones due to Class B, have been renderednegligible. This circuit has the potential to beblameless, but achieving this depends on carein cabling and layout. It does not aim to be acookbook project; for example, overcurrentand DC offset protection are omitted.

The investigation presented in parts 4 and 5concluded that power fets were expensive,inefficient and non linear. Bipolars make goodoutput devices. The best BJT configurationswere the emitter follower type II, with leastoutput switch -off distortion, and the comple-mentary feedback pair (CFP) giving best basiclinearity and quiescent stability.

I assume that domestic ambient tempera-tures will be benign, and the duty moderate, sothat adequate quiescent stability can beattained by suitable heatsinking and thermalcompensation. The configuration chosen istherefore emitter follower type II, which hasthe advantage of reducing switch -off nonlin-earities (Distortion 3c) due to the action of R15in reverse biasing the output base emitter junc-tions as they turn off. The disadvantage is thatquiescent stability is worse than for the CFPoutput topology, as there is no local feedbackloop to servo out Vbe variations in the hot out-put devices.

The NFB factor was chosen as 30dB at20kHz, which should give generous HF sta-bility margins. The input stage (current sourceTr1,Tr14 and differential pair Tr2,3) is heavilydegenerated by R2,R3 to delay the onset ofthird harmonic Distortion 1. To assist this thecontribution of transistor internal re variationis minimised by using the unusually high tailcurrent of 4mA. Trio,ii form a degeneratedcurrent mirror that enforces accurate balanceof the Tr2,3 collector currents, preventing sec-ond harmonic distortion. Tail source Tr] ,14 hasa basic PSRR 10dB better than the usual twodiode version, though this is academic whenC11 is fitted.

Input resistor R1 and feedback arm R8 aremade equal and kept as low as possible con-sistent with a reasonably high inputimpedance, so that base current mismatchcaused by beta variations will give a minimalDC offset. This does not affect Tr2-Tr3 Vbemismatches, which appear directly at the out-put, but these are much smaller than theeffects of lb. Even if Tr2,3 are high voltagetypes with low beta, the output offset shouldbe within ±50mV, which should be quite ade-quate, and eliminates balance presets and DCservos. A low value for Rs also gives a low

THD+N(%) vs FREQ(Hz) 18 NOV 93 15:38:230.2

0-1

0.010

o.00l

::::-.-1

-:

----- 1--,--

J

i

----- 1-----r-.1

J

fl241.....410k=.0..aPk.S414_,

,

:::::

' 1-'

.

.

:::-1-3..lit.___. :::::,

1

L 1..

1

,-.--1,r

p

l L..l ...1..Ji..L

l L'I'l1 1

T".1,T,--L1J1.

.

.

.

.

1-11

......________LL

111J . -------

p ,

1'

1 plp

'1.0

...,..

...

.....

--r-,-r,--

-t-',:3_-':-::::,-L-6.1J--AJ

L -1---

.

L ,

: '

,-1-,1J-J

L

,

4

1 p

1

r-r,-,-.,

p

hi

H

- -

....

::::: t::1:1:.L..::L.:::._::::,::,,,=:,,:,p

I .

1. -- 1.- -1--1-.:-

1. 1

.. 1 .

1 . I

--'. E --4--i-

j$ '

L '

,-, r--,--,,--1-

..

t _.r

-1-, L I

1 .

NOrtnal

Jr

1-1--1-11

J 1

-

-

'Th_r

....

1..

1.

1-1

,r,pl

::,.1.

L:

'

.'r

i--J---

2 -poleT

1 4--J

--

_J_-L._

--1-

..j.

JJ

Ape

--1-

---

I-

.

_

-

1_

-1- -r-,-r77.

7, . ... -,-r, r 7-7-7--rrr

' ' " , . , . . ...000510 100 1k 10k 50k

Fig. 8. Class B amplifier with simple quasi -complementary output. Lower trace is for two -polecompensation (80kHz bandwidth.

THD+N(%) vs FREQ(Hz) 18 NOV 93 15:49:050.2

0 1

0.010

0.001

----- 1

--- 1---.--r,-rrlrr----,

:I

----- J

-1

A

1

----- ,-----r--

'J

S.W.46....frfVFPE.1%-4.L.

-----

-----

, LJ1. I.

,--1-14-1J.L 1.J

---t1J.it, ,

.

11-.1-;±1.--1JJ1J

- HlTi.,--L1.11,

;-4-14

1

,

.

-' t*

...

-t.1.1..o.

:A. --------

-------,

I 1 ,

..

--------

I;

.f

ni

1

---r7r7--,0

,,,--L-,J-1--11--i

p --A-,...

L-

h 7

t.._.

:

1- 1.-1-1J-JJ.

;-

1

-

.

$

L1 t.

-H---Hr p 1

-..

.

-I-

1

L1_,-L!,--.1-:

,-,---)

- -;

, t

. .

..

. e

J

,:--1-J,-.1--

1

J

---4j, .--4-1-4-----. .

I .

---1,

19 0 ii

..:-.:1c, 4

----- L--i-J-

:-

,--

1 -; T

lip1

-- ---- L_ r

L -1-J-1

L J-1--

r p

. .

.

M 0 (

; '

r j

4 J.,

. .

it 7

. .

j T

T

--H-:

...

. e .

6.1.r1,- ---- i

r,r, T

. . .

1.1

0-,_ 4

LPL - --- r

-----

1..t

...

:-:' T

2 -pole...

rLr, T

rLr. -- 1-

-- 1--:.--.

------

,_Lr- a-:,.-_1____

_j__,_J

Ap.

e e

-

'$,

7 ,--r,-,-.74.

-r--r,,,. .. . . . . .

r -7-.. , . , .

-rrr$. ,1..0005

10 100 1k 10k 50k

Fig. 9. Class B amplifier with quasi -comp plus Baxandall diode output. Lower trace is the 2 -pole case(80kHz bandwidth).

value for R9, which improves the noise per-formance.

The value of C2 shown (220µF) gives an LFroll off with R9 that is -3dB at 1.4Hz. The aimis not an unreasonably extended sub -bassresponse, but to prevent an LF rise in distor-tion due to capacitor non linearity.

For example, 100pF degraded the THD at10Hz from less than 0.0006% to 0.0011%.Band limiting should be done earlier, with nonelectrolytic capacitors. Protection diode D1prevents damage to C2 if the amplifier suffersa fault that makes it saturate negatively; itlooks unlikely but causes no measurable dis-

tortion I°. C7 provides some stabilising phaseadvance and limits the closed loop bandwidth;R20 prevents it upsetting Tr3.

The vas stage is enhanced by an emitter fol-lower inside the Miller compensation loop, sothat the local NFB which linearises the vas isincreased by augmenting total vas beta, ratherthan by increasing the collector impedance bycascoding. The extra local NFB effectivelyeliminates vas nonlinearity (Distortion 2).

Increasing vas beta like this presents a muchlower collector impedance than a cascodestage due to the greater local feedback. Theimprovement appears to make a vas buffer to


AUDIO

0.2THD+N(%) vs FREQ(Hz) 18 NOV 93 17:06:47

.

.

.

.

. , .

..,

,

. ,

.. .

,

.

.. ,

.. Ap.: I I :: :

. $

: : : .

0:1 -----t.:.71321--------::1:"-------:r:':;:c:::: ..-_L_4-L_-'-0,

L_: ----- I --::::::-J--,--11.1, -/-,.4--1----- J ----- L --11-11A -------- L-4-1.-:---/-: :.--i-J-4--F.r1,- ---- 4 -- J----

1

1r : 1 'I : 1 T T r, r. ;

1:t L :. 1

1. - ' I 1

. J 1-

. . . ...1

.1 . .. ,

0.010 j.i hf-2jii1----:-1 LIIT-----i.C: ___ ---t------ ----J $.. a ...1.1 L . L. , . L . -1 . L.1 . J _

J: 11-4-1:4.. : T -4j-i-: ----- --4-14 :: _ J .

1J

r1-

T,1,7L1JILr '-r,--,-1-6.1a-AJi0 F ,-1-,--,r,,..--1-4-1-1. 4_, -_

s s : . 1 . 1 ,11 . 1 . 1.1

1----- 1--,--.

__!..; 4,.. ---------- ;... 4___...; ---- _ -- ;._i_T__ !:__ _T__ ___.

. I

.I .

----- --r- - ,-- $-: r :----: :

:..

,::.. :

. $

..., $ .

. . . $ . $ .

0.001 ----- ... -- -1- -._ -.-ily.r.i ---- ---- c - T-.. -: j-- r -j--r-,-1--r--r.-:-' ----- T--;r).- ---- 1--J--:------J ----- I-- 1.1.111

T - T $ -.' r -1

: : I :

.0005, -.--.--,-

. , , , .

r --1---:-r-. , .

, , , ,

10 100 1k 10k 50k

Fig. 10. Class B amplifier with complementary feedback pair (CFP) output stage.

eliminate Distortion 4 (loading of vas collectorby the nonlinear input impedance of the outputstage) unnecessary. Cdom is relatively high at100pF, to swamp transistor internal capaci-tances and circuit strays, and make the designpredictable. The slew rate calculates as40V4tsec. The vas collector load is a standardcurrent source, to avoid the uncertainties ofbootstrapping.

Quiescent current stabilitySince almost all the THD from a blamelessamplifier is crossover, keeping the quiescentoptimal is essential. Quiescent stabilityrequires the bias generator to cancel out theVbe variations of four junctions in series; thoseof two drivers and two output devices. Biasgenerator Tr13 is the standard Vbe multiplier,modified to make its voltage more stableagainst variations in the current through it.These occur because the biasing of Tr5 doesnot completely reject rail variations: its outputcurrent drifts initially due to heating thuschanging its V. Keeping a Class B quiescentstable is hard enough at the best of times, andso it makes sense to keep these extra factorsout of the equation.

The basic Vbe multiplier has an incrementalresistance of about 2052; in other words itsvoltage changes by ImV for a 50µA drift instanding current. Adding R14 converts this to agently peaking characteristic that can be madeperfectly flat at one chosen current; see Fig. 5.Setting R14 to 220 makes the voltage peak at6mA, and standing current now must deviatefrom this value by more than 500p.A for aImV bias change. The R14 value needs to bealtered if Tr5 is run at a different current. Forexample, 160 makes the voltage peak at 8mAinstead. If TO3 outputs are used, the bias gen-erator should be in contact with the top or canof one of the output devices, rather than the

heatsink, as this is the fastest and least atten-uated source for thermal feedback.

Output stageThe output stage is a standard double emitterfollower apart from the connection of R15

between the driver emitters without connec-tion to the output rail. This gives quicker andcleaner switch -off of the outputs at high fre-quencies; this may be significant from 10kHzupwards dependent on transistor type. Speedup capacitor C5 improves the switch -offaction. C6, R18 form the Zobel network whileL1, damped by R19, isolates the amplifier fromload capacitance.

Figure 6 shows the 50W/852 distortion per-formance, about 0.001% at IkHz, and 0.006%at 10kHz. The measurement bandwidth makesa big difference to the appearance, becausewhat little distortion is present is crossoverderived, and so high order. It rises at6dB/octave, the rate at which feedback factorfalls. The crossover glitches emerge from thenoise, like Grendel from the marsh, as the testfrequency increases above 1kHz. There is noprecipitous THD rise in the ultrasonic region,and so I suppose this might be called a highspeed amplifier.

Note that the zigzags on the LF end of theplot are measurement artifacts, apparentlycaused by the Audio Precision system tryingto winkle distortion from visually pure whitenoise. Below 700Hz the residual was purenoise with a level equivalent to approx0.0006% at 30kHz bandwidth. The actualTHD here must be microscopic.

This performance can only be obtained if allseven of the distortion mechanisms are prop-erly addressed; Distortions 1-4 are determinedby the circuit design, but the remaining threedepend critically on physical layout andgrounding topology.

Figure 7 shows the startling results ofapplying 2 -pole compensation to the amplifi-er. C3 remains 100pF, while Cpl was 220pFand R Ika The extra NFB does its workextremely well, the 10kHz THD dropping to0.0015%, while the 1kHz figure can only beguessed at. There were no unusual signs ofinstability on the bench, but I have not tried awide range of loads.

This experimental amplifier was rebuilt withthree alternative output stages: the simplequasi -complementary, the quasi-Baxandalland the CFP. The results for both single andtwo pole compensation are shown in Figs 8, 9,and 10. The simple quasi comp generatesmore crossover distortion, as expected, and thequasi Baxandall version is not a lot better, dueto remaining asymmetry around the crossoverregion. The CFP gives even lower distortionthan the original EF-II output. Figure 10shows only the result for single pole compen-sation; in this case the improvement with twopole was marginal and the trace is omitted forclarity.

The AP plots in earlier parts of this series weremostly done with an amplifier similar to Fig. 6,though of higher power. Main differences werethe use of a cascode-vas with a buffer, and aCFP output to minimise distracting quiescentvariations. Measurements at powers above100W/8L1 used a version with two paralleledoutput devices.

Next month: In the final part of this series,Douglas Self presents a full Class A amplifierdesign.

References1. Otala, An Audio Power Amplifier for UltimateQuality Requirements. IEEE Trans on Audio &Electroacoustics, Vol AU -21, No.6, Dec 1973.2. Baxandall, Audio Power Amplifier Design.Part 4. Wireless World, July 1978, p76.3. Takahashi, et al, Design & Construction ofHigh Slew -Rate Amplifiers. AES 60thConvention, Preprint No. 1348 (A-4) 1978.4. Self, Crossover Distortion & Compensation.Letters, Electronics & Wireless World, Aug 1992,p. 657.5. Widlar, A Monolithic Power Op -Amp. IEEEJournal of Solid -State Circuits, Vol 23, No 2, April19886. Bonello, Advanced Negative Feedback Designfor High Performance Amplifiers. AES 67thConvention, Preprint No. 1706 (D-5) 1980.7. Pernici, A CMOS Low -Distortion Amplifierwith Double -Nested et al Miller Compensation.IEEE J. Solid -State Circuits, July 1993, p. 758.8. Fast Compensation Extends Power BandwidthLinear Brief 4, National Semiconductor LinearApplications Handbook 1991.9. Feucht, Handbook of Analog Circuit Design.Academic Press 1990, p. 264.10. Self, An Advanced Preamplifier. WirelessWorld, Nov 76, p. 43.


LIVELINESDual port serialcommunications for your PC

The PC 47,48 and 49AT boards eachprovide two independent serial portsfor any PC/XT/AT(ISA) computer.RS232, RS422 and RS485 standardsare supported and all combinationsare possible with this range of lowpower boards.

9 pin D connectors are provided forthe RS422 and 485 ports. RS232 canbe connected via 9 or 25 way Dconnectors and charge pump circuitryensures signal levels in excess of 7Vare transmitted.

Interrupt and base addresses areindependently selectable for eachport, full, half duplex and multidropcommunications are fully supportedand all boards use the industrystandard 82C450, UART.

Designed and manufactured byAmplicon Liveline each board issupplied with a comprehensivetechnical manual and interrupt drivendevice driver software is available.Write in number 1

750kHz professionaldata acquisition withWindows supportPC226 from Amplicon Livelineprovides 16 true differential 12 bitanalog inputs each with dynamicallyprogrammable gain. PC226 hassample rates up to 400kHzmultichannel and 750kHz singlechannel, programmable scanninghardware, flexible triggers and a 2048sample FIFO to ease programming inhigh speed applications.

PC 226 is supplied with menudriven software for DOS andWindows, LabTech Notebook driversand comprehensive well documentedlibraries for most popular high levellanguages.

Optional software support forMicrosoft Windows includes acomplete Dynamic Link Library(DLL) and two icon driven packages,Signal Centre for signal capture andanalysis and TRACS for processmonitoring, visualisation and control.Write in number 2

48 lines TTL I/O with 16bitcounter timersPC 14AT available from Amplicon Liveline is ahigh quality general purpose interface board forPC/XT/AT(ISA) computers.

The board has 48 lines of TTL compatibledigital I/O provided as six 8 bit ports four ofwhich can be programmed to be all inputs or alloutputs and two which can be split to be 4 inputsand 4 outputs if required. There are also three16bit counter timers which can be used togenerate and measure pulses at up to 5MHz.

An on board 4MHz crystal oscillator is alsoincluded on PC 14AT along with LEDs showingthe status of a selection of the I/O ports, thesecan be used as programming and debuggingaids.

PC 14AT is suitable for a wide variety ofinterface, monitoring and control applications

and an LP (low power) version PC 14LP isavailable for installation in laptop computers,both boards have flexible base address andinterrupt support. Write in number 3

TM E%Air II MI

Large colourayilalogese -free

24 hoarph andfax lines

Knowledgeableengineersfor help

&weedaydespatch

Excelleatsupportmaterial

CertifiedqualityFM 15094

Versalessupport

For up to the min_ite information onover 1,00n high quality, high specificationelectronic products, make sure you've gotthe latest AIMPLICON catalogue.

it's not just first class equipment thatmakes AMPLICON your first choice- we provideoutstanding sery ce too: like telesales engineersfor pre and post sales support. Easy to managecorporate accounts Excellent documentation...and same day despatch for ordersreceived before 5 pm.

To order your free :atalogue, simplyphone the winning line today.

AMPLICON014.0C

GET C111111- - 111177-11-1 4411/1124-ICCPItieDATA ACQUISITION POWER SUPPLIESINSTRUMENTATION INDUSTRIAL DATACOMMSTEST AND MEASUREMENT

ELECTRONIC ENGINEERS

REFERENCE BOOKThis reference book is divided into five parts:

techniques, physical pheromena, materials andcomponents; electronic design and applications. The sixthedition was updated throughout to take into accountchanges in standards and materials as well as advancesin techniques, and was expanded to include newchapters on surface mount technology, hardware andsoftware design techniques, semi -custom electronics anddata communications.

6th Edition

ELECTRONICSENGINEER'S

REFERENCEBOOK

\ -

Edited by F F Mazda

Fraidoon Mazdahas worked in theelectronics andtelecommunica-tions industry forover twenty years,and is currentlyProduct andOperationsManager, GenericNetworkManagement,with NorthernTelecom. He isthe author of sixtechnical books(translated intofour languages)and the editor oftheCommunicationsEngineersReference Bookpublished byButterworth -Heinemann.

CONTENTS:TechniquesTrigonometricfunctions andgeneral formulae;

Calculus; Series and transforms; Matrices and determinants;Electric circuit theory; Statistics. Physical PhenomenaQuantities and units; Electricity; Light; Radiation; The ionosphereand troposphere. Materials and components Resistivematerials and componerts; Dielectric materials and components;Magnetic materials; Inductors and transformers; Relays;Piezoelectric materials and components; Connectors; Printedcircuits; Power sources; Discrete semiconductors; Microwavesemiconductor devices; Optical digital integrated circuits; Linearintegrated circuits; Semiconductor memories; Microprocessors;Application -specific integrated circuits; Electron microscopy;Digital design; Software enginee-ing; Digital systems analysis;Control systems; Antennas and arrays; Noise management inelectronic hardware; Noise and communication; Computer aideddesign; Television and sound brcadcasting. ApplicationsCommunication satellites; Point -b -point communication; Fibre -optic communication; The integrated services digital network(ISDN); Local area networks; Radar systems; Computers andtheir application; Videotape recording; Office communications;Medical electronics.

1006 pages PAPERBACK ISBN 07506 0809 9£42.50 (inc post & packaging)

6th EditionF F Mazda

* Now available to Electronics World & Wireless World readersin paperback.* Expert coverage of all aspects of electronics* Over 50 contributors" For electronic engineers, technicians and students

Please supply me copies of the ELECTRONICENGINEERS REFERENCE BOOK (ISBN 07506 0809 9) @£42.50 (INC POST & PACKAGING)

Add VAT at local rateNB ZERO RA-E FOR UK & EIRE TOTAL

Business purchase: Please send me the books list-ed with an invoice. I will arrange for my company to pay theaccompanying invoice within 30 days. I will attach my businesscard/letterhead and have signed the form below.

Guarantee: If you are not completely satisfied, books may bereturned within 30 days in a resalable condition for a full refund.

Remittance enclosed £Cheques should be made payable to Reed Book Services Ltd.

Please debit my credit card as follows:

Access/Master Barclay/Visa

Amex Diners

Credit Card No. Exp date

NAME (Please print)

ORGANISATION

STREET

TOWN

COUNTY POST CODE

COUNTRY

DATE TELEPHONE NUMBER

SIGNATURE

T3000

VAT RATES6% Belgium, 25% Denmark, 5.5% France, 7% Germany, 4%Greece, 4% Italy, 3% Luxembourg, 6% Netherlands, 5%Portugal, 3% Spain. FOR COMPANIES REGISTERED FORVAT, PLEASE SUPPLY YOUR REGISTRATION NUMBERBELOW (customers outside the EEC should leave this partblank)VAT NO.

If in the UK please allow 28 days for delivery. All prices are correct attime of going to press but may be subject to change.

Please delete as appropriate. I do/do not wish to receive further detailsabout books, journals and information services.

Credit card ordersaccepted by phone.Call 081 652 3614.

Please return to: Lorraine Spindler, Room L333,Quadrant House, The Quadrant, Sutton, Surrey SM2 5AS

NEW THE DEFINITIVE ÒFF -AIR'FREQUENCY STANDARD

III2 40,7 kt

* Provides 10MHz, 5MHz 8 IMHz* Use 0 for calibrating equipment that relies on quartz crystals, TCX0s.

VXCOs, oven crystals* Phase locks tc DROITWICH (rubidium controlled and traceable to NPL)* For ADDED VALUE also phase locks to ALLOUIS (cesium controlled

and traceable to OP - French eq to NPL)* British designed and British manufactured* Now with Sine Wave Option, output 1 volt into 500

'' il:195.+WaOnly

carriage extra

Output frequencies -10MHz, 5MHz, 1MHzShort term stability - betterthan I a10 alt sec)Typical - 0 sec)Long term - tends to2010'12 (1000 sec

IDEAL BEGINNERS SCOPE, SCOPEX 4S6 6MHz SINGLE TRACE, INT/EXT TRIG, TRACELOCATE, BRIGHT LINE AUTO, 10mV SENSITIVITY, 1pS-100mS/cm, etc. i'.95

TELEOUIPMENT 061A 10MHZ DUAL TRACE FROM £99 7SEG 12" 8 9" DISPLAYS DIGITEKSIGNALEX £15 8 £10TELEOUIPMENT 562 SINGLE TRACE 5MHz SCOPE C95 RADIOMETER BKF6 DIST. METER 20HZ-22KI-2 POAIWATSU SS5116 C UAL TRACE, 10MHZ £175 RADIOMETER AFM2 MOD. METER 7MHZ- GFZ POATELEOUIPMENT D1011 10MHZ DUAL TRACE £165 UPS MICROPAC SX250 250W £135IWATSU SS -5802 DIGITAL STORAGE £295 CITOH CX6000 6PEN A4 PLOTTER. CENTAS232 £135TEKTRONIX 2215 60MHZ 2 TRACE DEL T/B £450 GENERAL RADIO 1531A XENON STROBCTAC £79TEKTRONIX 453A 50MHZ 2 TRACE DEL TIB £249 THANDAR TG501 FL NC GEN 0.005HZ-5M-IZ £165H.P. 8405A VECTOR V/METER 1GHz £395 BACHARACH MV2 MERCURY SNIFFERS £69PLESSEY TCTIO SIG GEN/ANAL 50-300 BDS £95TEK 465B 100MHZ DUAL TRACE DEL T/B C475TELEOUIPMENT D67A 25MHZ, 2T, DEL T/B £215H P 1700A 35MHZ DUAL TRACE £249HITACHI VC 6015 10MHZ DIGITAL STORAGE £295HP1340A X -Y DISPLAYS 0149OERTLING V20 StIGLE PAN BAL 0 1MG/200GM £69ANALYTICAL BALANCES WITH WEIGHTS 250GM £69VACUUM PUMPS 1.5 8 2.8 CU M/HR £125 & £149KINGSHILL NS1540 15V 40A PSU's CASED, AS NEW £195ACRON 402P SYNCHRONISING PULSE GEN & 605PENCODER £375ea £695/prMARCONI TF230c AM/FM MOD METER PRTBLE £249MARCONI TF2330 WAVE ANALYSER 20Hz-50kHz £149H.P.5315A 1GHz 0/CTR, OPTS 1. 2 0 3 C595LEVELL TM6B MICRO V -METER 450MHZ £95LEVELL TM3B MICRO V -METER 3MHZ C85

FISONS F I -MONITORS LIO LEVEL SENSCRS £85JAOUET TIMER I 1 ICSECS. RES'N OISEC £59COMMODORE PETE, D/DRIVES. PRINTERS £29EACOMARC 2303 MV SOURCES, DUAL RAFLIE C49INTRON IFG422 FUNC GEN 0.11-12.2MHZ 0125ECG MONITOR SEM430 WITH SEM420/2 C75GOULD 2400 4 -PEN "HART RECORDER £149kV PLOTTERS A3 & A4 RON £35 to £139PHILIPS PM6456 FM STEREO GENERATOR £195VALVE TESTERS AVO MK 1,2.3 frcm £49 to £89MARCONI TF2300 FM/AM MODULATION METER £195McKENZIE 7 DAY TEMP/HUMIDITY RECORDER £95FEEDBACK SS0603 1MHz SINE/SO OSC £125FARNELL E350 0-353v 100mA, 2 a 6.3V £59 to £69COMMUNICATIONS RECEIVERS. HF, LF,VHF POALCR MARCONI TF1213 0.25% £95LCR COMPONENT COMPARATOR AVO CZ4E7/S £95

LIST AVAILABLE BUT 1000's OF UNLISTED BARGAINS FOR CALLERS. ALL PRICES EXC. OF P.P AND VAT

QUALITY ELECTRONIC EQUIPMENT ALWAYS WANTED

.1%. HALCYON ELECTRONICS423, KINGSTON ROAD, WIMBLEDON CHASE, LONDON SW20 8JRSHOP HOURS 9-5.30 MON-SAT. TEL 081-542 6383. FAX 081-542 0340


(PROMulator

/ Emulates up to 41 Mbit EPROMsvia one standardprinter port

/ Downloads 27256In 3 seconds

Accepts Intel Hex,Motorola S -Recordsand Binary files

Tel: 081-441 3890Fax: 081-441 1843

ROM Emulator

FastFlexible

/ Emulates 24, 28, 32,40 and 42 pin devices

/ Full screen editor

/ Other models availableup to 8 Megabits andwith bi-directionalcommunications

CALL FORFULL DATA SHEET

e Pk

13011ANIUNICATIM3


SYSTEM 200 DEVICE PROGRAMMER

SYSTEM: Programs 24,28,32 pin EPROMS, EEPROMS,FLASH and Emulators as standard,quickly, reliably and at low cost.

Expandable to cover virtually any pro-grammable part including serial E2, PALS,GALS, EPLD's and microcontrollers from allmanufacturers.

DESIGN: Not a plug in card but connecting to the PCserial or parallel port; it comes completewith powerful yet easy to control software,cable and manual.

SUPPORT: UK design, manufacture and support. Sameday dispatch, 12 month warranty. 10 daymoney back guarantee.

m 71.4=NM

MOP ELECTRONICS Ltd.Park Road Centre

Malmesbury, Wiltshire. SN16 OBX. UKTEL. 0666 825146 FAX. 0666 825141

ASK FOR FREEINFORMATION

PACKGERMANY 089/4602071NORWAY 0702-17890ITALY 02 92 10 35 54FRANCE /1)69.30.13.79Also from VEROSPEED UK

Field Electric Ltd. Te: 0438-353781 Fax: 0438 359397Unit 2, Marymead Workshops,Willows Link, Stevenage, Herts, SG2 8AB.

IBM & Compaq colour 14" VGA monitors from £140 + C/P16 bit VGA controller card £18 C/P £3.751.44Mb Laptop floppy desk drives new £16 C/P £316 bit Ram cards with Ram from £25 C/P £3PS2 upgrade cards please ring for list72 pin Simms for PS2 Compaq Dell etcParalle & Serial data switches up to 8 ways Ring for detailsSony colour monitors with data 9" screen £130 C/P £17 240VAC input casedCompaq enhanced at keyboards (no cable) £11 C/P £55Mb Sipp Motherboard with various chipsMC68451L10: MC68010FN10: SRM2264C12 etc: £60 C/P £6Wyse 60A Terminals with ASC1 keyboards, new & boxed 240VAC £190 C/P £25PC power supplies up to 600w, in stock Prices start from £19 C/P £6IBM 3363 optical disk drive £100 C/P £20

TEST AND MEASUREMENT EQUIPMENTHP 18180A Protocol Analyzere/w manual etc: RS232C V24/RS449 £1250HP86A Computer New & Boxed £70Tektronix 454 150MHz 24ns rise time scope £250Tektronix 455 50MHz scope £250Tektronix TM504 M/Frame c/with DD501/WR501 LA 501 £90Tektronix 434 25MHz scope £95Tektronix 7B92/7A181\I Plug ins £95Telequipment D75 50MHz c/with V4 & S2A Plug ins £130Gould 0S1100A 30MHz scope (faulty HT) £75Lambda LES-F-03-OV-V 0-36v 24A Digital PSU £375Lambda LK 342A FMV 0-36v 52A £120AVO RM 215F3 AC Breakdown Tester £95Schlumberger 4000 Precision Sig: Gen: £150Schlumberger 4900 RF-AF Measuring Unit £150RFL 912 Gaussmeter £120Wyse 60A Terminals new & boxed with keyboards £195Sony Videotex Terminal KTX 1000Datatruck 7 short haul multiplexerGandalf GLM 518 Line miser

All above equipment + 17.5% VAT Please ring for c/p prices

We wound like the opportunity to tender for surplus equipmentES UMWOfficial orders, credit card telephone orders accepted with

Access, Amex, Diners, Visa cards. Overseas enquiries welcomecip rates UK mainland only. Please ring for c/p rates not shown

All prices inc VAT unless stated. Stock list available


February 1994 ELECTRONICS WORLD+WIRELESS WORLD


145

£100WINNER

CIRCUIT IDEAS SEND YOUR CIRCUIT IDEAS TO THE EDITOR, ELECTRONICS WORLD,

QUADRANT HOUSE, THE QUADRANT, SUTTON, SURREY SM2 5AS

DO YOU HAVE A f100 CIRCUIT? EACH MONTH'S TOP CIRCUIT IDEA AUTHOR WILLRECEIVE f100. ALL OTHER PUBLISHED IDEAS WILL BE WORTH f25. WE ARE LOOKING FORINGENUITY AND ORIGINALITY IN THE USE OF MODERN COMPONENTS

Smart fet battery chargerThis circuit uses an external transistorwrap -around to boost the current

capability of a voltage regulator for aconstant voltage lead -acid battery charging

application. Using theInternational Rectifier

1RFS3010 smart mosfet(EW+WW Dec 93 p990) inthis position conferscurrent, voltage andthermal protection to the

circuit. Standard devices willalso work but will self-destruct

under short circuit and other unfavourableconditions.

The normal configuration would use apositive variable regulator of the 317 typetogether with a pnp bypass. Since the IR fet

is an n -channel device, it requires the use ofa negative rail regulator producing anegative rail but this shouldn't be problem ifthe raw DC supply is made floating.

Operation is self explanatory. The cut -inpoint for the external fet will be determinedby the value of the input resistor to thevariable regulator. It should be chosen sothat the voltage across it fully enhances thechannel before the 337 reaches its own 1.9Acurrent limit. The DC supply should haveenough overvoltage to allow for this.Although the devices have inherent thermalprotection, they require appropriateheatsinking for continuous operation.Nick WheelerSuttonSurrey

10MHz/1 MHz marker generatorA common -base amplifier with a diode

.connected inversely across base andemitter and fed with TTL input producesnarrow negative -going pulses at thecollector. Two such devices fed with10MHz and 1MHz input generate markerpulses - in my case for a 300MHzspectrum analyser.Signal from a 10MHz oscillator goes tothe 10MHz pulse generator directly and,by way of a decade divider, to the 1MHzgenerator. Depending on the position ofthe power switch, the voltage supply goeseither to the 10MHz circuit alone or toboth circuits to produce a 10MHz/1MHz"comb".Resistor R1 isolates the load, causing a20dB loss. The 10MHz output is-50dBm, that at 1MHz being -70dBm.D HutchinsonBromsgroveWorcestershire

-20V 4R7

10W1

RSF3010

LM337-1.2-14V

220

2k5

1t.

78L05

+5

Common -base amplifiers produce fast, narrow pulse when fed with TTLedges to form a spectrum analyser marker generator at 1MHz and 10MHz.

100 n

10 MHzmodule

Jr

100 n

3 - 20 pCl

IN 914

31

470 0

50 0 load

R1

470

2 5

10n 220 10 n

2 N 2369 A

00

10 p

IN 914

74 HC 390

390

SNM1 5 k

10 MHz

+12v*-n 60 mA

10/1 MHzSPCOcentre off


CIRCUIT IDEAS

A -to -D card for PCsn EC Plessey's ZN448 has a built-inV clock running at up to 1MHz, has itsown reference and is an 8 -bit device,resolving to 39mV. This card converts a±5V input to 256 bits, 128 bitscorresponding to OV.

Since the ZN448 is a single -channeldesign, a 74HC4351 8 -channel analoguemultiplexer precedes it, the output on Zbeing selected by Y0_7 when LE latches theinput -select data on S0_2. The analoguevoltage at Z goes to the analogue input ofthe ZN448 via the potentiometer chain.

Conversion starts when WR\ goes low,RD\ enabling the data latch, valid dataappearing 8-9 clock cycles after theconversion has started.

When the address data on the P inputs ofthe 74LS688 comparator corresponds to thaton the 8 -way dip -switch, the P+Q\ outputenables the 74LS245 bidirectional bus bufferand gates read and write signals to theZN448. IOR\ also controls the direction ofthe buffer for read and write.

Addresses must be in the 300-31F range.For 300, switch 1 is off (A8_9 high), 2 and 4on (A5,7 low) and setting the rest on altersthe address upwards. Base address plus 1enables the multiplexer latch.

To calibrate the card, apply +5V and

A9 EAS E>

A7A6A5A4A3A2Al

USA

741.508

3

46

adjust VR2 until D7 flicks between 0 and 1 togive a reading on the screen of 254-255.Now apply -5V and adjust VR1 for a flickerbetween 0 and 1 on D0, the screen showing0-1. The adjustments are interactive.

This program continuously reads theZN448 and displays the result on screen.

10 basE=4300:basel=#301 rem;for example!

20 out basel,X rem; select anyinput channel

30 out base,Y rem; write toZN448 to start

40 for t=1 to 50:next T rem;necessary

50 A=inp(base) rem; collectresult

60 print A rem; display result70 goto 20 rem; start again80 rem; X=0-7: Y=any value

Mel SaundersLeicester

A -to -D converter on a PC card resolves to39mV. Configuration shown measures ±5Vinputs

1.13

11

13

1517

Ar171 [> lo

AO

DODID203D405DeD7

POP1P2P3P4

P6P7

0001

020304Q5Q607

R1

SILRES9

10

9

ID

to

D1PSW -9

D19-

74LS688

FEW ED>

19

2

U2 74 L 532

7OR

U6B

74 L SOS

6

0

DATpo,cin

DATAOD /-i- AlDA TA01 /"^, DATADOA2 92

5 A3 93 I glitE gITH4

A4 B4 .

A7 B7 DAY

9 A6 99 1 DA TAD

A5AO

7 4L S 24 5

74L S32

Ul

011Th

RN MTDDDt

D2 Amm

D4

06 WW1D7 VRE FD

RE X T

MI

f

SIDE A.J2 1

J2

J2

J2 II>

42 13>

J2 15>

J2-1

47

.12 21

C3OL K J2 2

J2-25>

J2-2

.12-3>

J2 3.

t(7) ow .12

C:11-6TY J2 3>

<-11ENR J2-4.

-CJIE1114 J2 4>

'CD t12y J2 45

J2 4

Ca -12V J2 41>

J2-5>

-Q-55 J2

J255>

(ITIITErr J2 50

J2 -6>-

C3/.0

<= Al

CJA2

CDA3

CjA4

C3A5

C.) AS

.C3 A7

<=1A8

<= A10

'D All

Al2

Al3

1=J Al4

<=1 A15

At6

<=1 A17

(=I All

<=I A19

C1A7ri

<=I DO

CD DI

<=02

Q D3

<= D4

<=ID5

<= Da

C3 D7

DAUM)DA 1 AG IDAYA02

29448

R3

715

- C1 ±- C2T10007- 4.7a -5V

VR1

51(

4`1 VR2

5

111

4

U-55

U4

SO YOV YI52 Y2E2 Y3El Y4LE Y52 Y$SEE Y7

17

4HC4351

.11-1

J1 2

J1

J1 4

JI 5

J1 6

.11 7

.11

J1 9


CIRCUIT IDEAS

28

58

Metal strip 5 x 1 mm

Diodee.g. Siemens BAR 19

1 nF capacitor

Fig.1. Dipole dimensioned for the microwave heating frequency, 2.45GHz; the diode is a Schottky type.A wire dipole is somewhat subject to static charges, which are relatively unimportant in the slot antenna.

0.5 mmmetal plate

or box

3 -5 mmslot

Diode

nFcapacitor

Radiation detectorMicrowave heating in domestic ovensand in industrial processes must

comply with radiation standards. Newequipment does so, at least when properlyloaded, but older ovens can leak throughdeteriorating door and service openings,particularly when improper loadinggenerates harmonics of the magnetron's2.45GHz. When calibrated, this detectormeasures 0.01-10mW/cm2 power density.

Figure 1 shows the simplest type ofdetector -a simple, half -wave dipole witha Schottky diode in the gap, of correctdimensions for 2.45GHz, although theseare not critical. Depending on the diodeand meter used, sensitivity is 1 to

-0

DC amplifier

10mW/cm2. The slot antenna is animprovement, being as sensitive to RFpower, but much less vulnerable to staticcharges. Adjust the distance from the ovenand the angle so that a maximum can beseen. If the meter shows full-scale at lm, itis showing up to 1W/cm2 and that isdangerous.

Adding an amplifier, as in Fig. 2,improves sensitivity to around0.01mW/cm2. Ceramic "radar" diodeswork as well as Schottkys, whereas glassseal diodes have their problems.

A more comprehensive circuit, shown asFig. 3, is provided with two detectors, onehorizontal and the other vertical, each with

68k 5,12

50k

5k

M10

Gain

6 K8

M1 P3

47k 10k

P1 P2 M1 215k

68k M2

47k

6 K8

t 10k

Comarators 0, 10, 20, 30 dB

its own amplifier. Comparators, set toproduce an output when preset levels areexceeded, activate the audio oscillatoralarm and trip the relay supplying power tothe magnetron.Ji i PolivkaMexican National AutonomousUniversityMexico

Level

P4 M1

10k

330

P5 M1 330WA

Fig.3. Complete detector circuit, including presetcomparators and an alarm circuit. The relay can beused to disconnect the power supply from themagnetron if radiation leakage becomes too high.

mains

P6 M1

P7 M1

T3

330

WA14k7

330

22k

Gain

Fig.2. Adding a meter (a cassetterecorder level meter was used inthe prototype) increases sensitivityto 0.01mW/cm2.

Alarmtone generator

22k

6k8

4k7 T2

50i2

TR1

1_<>

switched circuit6 - 24 V -

0

O

1 k5

D3 8 V stab. to IC's Trip -offrelay


CIRCUIT IDEAS

Single pot tunes Wien oscillatorcrying one of the resistors in a Wien bridge alters the frequency and

V alsc the attenuation. In the circuit shown here, the ailing resistor alsovaries tl-e gain of a compensating amplifier to compensate exactly for thevarying attenuation.

Since the inverting input of /C2 is a virtual earth, the arenuation of thebridge is determined by the setting of R6, which is also the nput arm ofthe compensating amplifier feedback network, the gain of which is nowR5/R6. Since the frequency varies as 1/ (R3,R6), R6 must hae a resistancerange of 100:1, increasing the amplifier gain in the same proportion.

Diodes D1,2 and R1 form the AGC circuit. As the amplitude of outputincreases towards the distortion region, The diodes begin to conduct onpeaks, t ringing R1 in to circuit in parallel with R2 which, with R4, sets thegain of :he maintaining amplifier /CI. Gain thereby reduces and amplitudestabilises. This function is usually performed by a thermistor or small lightbulb MTh, perhaps, a little less distortion but with a certain amount of"bounce".

An upper limit to the frequency is fixed by the gain of /C2 falling athigher frequencies - when more gain is needed for comrensation.W A CambridgeRichmondSurrey

Wien -bridge oscillator with c_ne variable resistor,which copes with varying b-ioge attenuation by

adjusting compensating amplifier gain.

Inductively isolated data linkI aductive coupling between two small

chokes up to 6mm apart has the advantageover optical coupling that the link can bemade across an opaque barrier, such asthrough the wall of a sealed plastic case.Inductance values shown here work for1200baud transmission, but 9600baudshould be possible with smaller chokes.

Complementary emitter -followers bufferthe input and drive the overdamped LCRcircuit R2CILI, in which short current pulsesflow at transitions without causing anybaseline shift in non -return -to -zero data.Inductor L. placed in line across the barrier,must be sensed not to invert the data.

Two comparators see the induced voltageacross L2 and produce low pulses forpositive and negative data transitions, R7eliminating ringing. These pulses trigger and

C?10n

retrigger the flip-flop, made from 4012gates, to reconstitute the data. During breaksin data, C4 charges and forces the thirdcomparator's output low, resetting the flip-flop to a known state, which is Mark for

1kWO.

Input

Tr1BC547

BC557Tr2

RS232. Resistor R16 and C5 do the same atswitch -on.S/ Kearley(Would Mr Kearley please contact oureditorial office? - ed)

INI43i48

R21k

-K D21N4148

'Cl470n

r

6mm--I..

Cho kes ononsame axis

L1

33mH

R9 Rio Rti

10mHL2

r

10n

R4680

10k

10k

R3

39k

R5

680

5

6

10n

0

22k

22kR8

t 3M370.

+5

Inductive link using twosmall chokes serves tocarry data through anopaque barrier, such asthe wall of an instrumentcase.

0v

2 D3

N

22kt 1M t IN4148R14 S' R15.7. D5

1/4 LP339

1/4 LP339

12

IN

4148

D41/4 LP339

1010-

11

R12 C4330k. T 220n

13

R13WIS3M3 C5

T100n

14/23 6145 ,ICD 4012

9101112

13

0 o/p

0. 0V


WE HAVE THE WIDEST CHOICE OFUSED OSCILLOSCOPES IN THE COUNTRY

CHASE HER 2000 INTERFERENCE MEASURING RECEIVER WITH MAINS

NETWORK .. ..01750

TEKTRONIX 7000 SERIES OSCILLOSCOPESAVAILABLE FROM £200. PLUG -INS SOLD

SEPARATELY

OATRON 1061A - 61/2 digit True RMS AC/Current . . .01250DATRON 1065 Multimeter 51/2 digit AC/DC/Ohms. IEEE. £600

HEWLETT PACKARD 34908 Bench Multimeter 51/2 digit 41,D1Ohms . L200

PHILIPS PAI2534 Multifunction IMIM 61/2 digit. GPIBtlEEE . . £450

MARCONI Digital Frequency Meter 24304 10Hz-80MHz . . L125

MARCONI Digital Frequency Meter 24313 1011,2011M14 . . 0150

MARCONI 11+10141 Counter Tuner 24370C 100Mlit.... ... E175

MARCONI Universal Counter Timer 2438DC-520MHz 0225

HP 331I8 Function Generator 0 1111-1MHz Sine/Sq/In E125

FEEDBACK FG600 Sine/Sqrtn 0.01116100KH1 . ................... HOMULTIMETERS Hand Held M2355-32 ranges AC DC 10 Amps Diode/

Transistor Tester. Freq counter E32.50

TEKTRONIX 2235 Dual Trace 100MHz Delay Sweep EAOO

PHILLIPS 3065 2+1 Channels 100MHZ Dual TB Delay 5,00 0100

TEKTRONIX 475 Dual Trace 200 MHz Delay Sweep F550TEKTRONIX 465 Dual Trace 200 MHz Delay Sweep E450

H.P. 17254 Dual Trace 275MHz Delay Sweep 0550

H.P. 17154 Dual Trace 200MHz Delay Sweep ...... £500

NAME& 605 Dual Trace 60MHz Delay £400PHILLIPS PM3217 Dual Trace 50MHz Delay Sweep ..... £400

PHILLIPS P5113244 4 Trace 50101110 Delay Sweep . £350

TELEQU1PMENT D63 Dual Trace 50MHz Delay Sweep LOBIWATSU 555/05 3 Channel 40MHz Delay Sweep L400

HITACHI 6122 Dual Trace 40MHZ ..... 0325

KIK115111 5530 Dual Trace 35MHz L180

000L0 051100 Dual Trace 30MHz £160

HITACHI 0223 Dual Trace 20MHz Deal Sweep __owHAMEL 203 6 Dual Trace 20MHz Component Tesler .. E260

£241000300 Dual irate .... . gmGOULD 00250E1 Dual Trace 15MHz ... ...... ....... ... .. _.0125TEKTRONIX 2430 Dual Trace 150MHz Digital Storage. spa*TEKTRONIX 466 Dual Trace 100MHz Delay Sweep Analogue Storage ...... 0450

H.P. 1/414 Dual Trace 100MHz Delay Sweep Analogue Storage_... ...0450THIS IS JUST A SAMPLE. NAP 01MER5 AVAILABLE

FARRELL ELECTRONIC LOAD R51030-35 lhw 30Amp 35 Volt E600

HP 8590 Sweep Osc with 86974 Plug-in. 26 5 400115 E300

RACALLOANA RF Power Meter 9104 0400

RACAUDANA 9341 Databodge. Automata 0 C.R 80 0350

WAYNE KERR 6905 Automatic Prens,n Bridge 005). 1900

WAYNE KERR B605 Automatic Component Bridge 0 I oh E350FRANCIS PSU. P/S70MK2. 70050/300/100..... 0300

FARRELL PSU H60/250 -60V. 0-25Amps Metered 0400

FARKELIPSUL301 0-30V. 0-5Amps Metered 080

FARRELL B30/200 -30V, 20Amps £250

FARRELL 030/100-300, 10Amps E200

H.P. 52098 0-3200.0 0 lAmps Metered.__.. 0125

MARCONI 122700 Universal LCR Bridge. Battery trorh 0150

AVO Characteristic MetertICM163 E300

FARRELL LA520 RF Power Amp. 1.5-520MHz 300mW E115

RACAL 9100 Absorption Wattmeter IMHz-IGH1 3W E100

PHIWPS PM5193 Programmable synthesizer/Function Generator

0 IMHz-50MH0 IEEE-488. As new £1500

MARCONI 2019 Synthesized AMIFM Sig Gen 80101,1040MHz MNMARCONI 2018 Synthesized AM/FM Sig Gen BOKHz-520MHz E950

EIPOANA 355D Microwave Frequency Counter 20110-18014 0950

RACAL 9921 Frequency Counter 3090 0300

RACALIANA 1991 Nanosecond Universal Counter 8800

RACAL 9002 Millivoltrneter True WS 101tHEI 5GHz E450

RACAL 93014 102 Millivolltrneter True RMS 104110-110110 2300

RACAL 9009 Automata Mod Meter 10MHz-I 56HZ Wide Deviation .. 0250

LYONS PG73N Pulse Gen. PRE IHz-20MHz. ...... ... El MCORD 1cmabon K50000 Logic Analyser 0503

FARRELL PS152011 Synthesised Sig Gen AM/FM 100104-520MHz 06110

LEADER L50216 AM/FM Sig Gen 0.1-3061Flz & 75- II 51.1114 0400

GRUEL L NICER Vibration Exciter System. Consisting of molter control 1047,PAS, Amp 2708 6 Exciter Body 480/ (Up to 1780N - 400109

EARNER Isolating TransformersG0506 240v 500VA. unused 050

NEW EQUIPMENTHAMEL OSCILLOSCOPE HM1005 Triple Trace 100MHz Delay Timebase.. MI

HAMEL OSCILLOSCOPE HM604 Dual Trace 60MHz Delay Sweep.._......... 0653

HAMEG OSCILLOSCOPE HM203. 7 Dual Trace 20MHz Component Tester 0362

HAMEL OSCILLOSCOPE HM205. 3 Dual Trace 20MHz Digital Storage.....0653

Al other models available- all oscilloscopes supplied with 2 probes

FUOI STAN Mumma* *EP all units E5)APOLLO 10-100MHz Counter Timer Ratio/Period/Time interval 0c 0222

APOLLO 100.100MHz (As above with more functions) 8325

METEOR 100 FREQUENCY COUNTER 100MHz .... CMMETEOR 600 FREQUENCY COUNTER 600MHz E145

METEOR 1000 FREQUENCY COUNTER 1GHz .. E189

WITH 500 FUNCTION GEN 0.280-5006X0 Sine5Sqlri EII9ORION COLOUR BAR GENERATOR Pal/P/Nideo E229

Al other Black Star Equipment available

OTHER BAK EQUIPMENT AVAILABLE

SPECTRUM ANALYSERSHP 1011 with 85556 Sir Plug-in 104111-181110 ... L2000HP 1411 with 85548 6 85528 500kHz-1150MHz E1300HP 140T with 85540 68552A 500kHz-1250MHz E1000HP 1411 with 8556A 6 85528 20110 3001411 E1000HP 140T with 85531.4 85520 IkHz-110PAHE 2800MARCONI TF2370 30Hz-110MHz ........... 01500HP 162C with 85580 I 00kHZ-15001010 . 01500HP 355/8/ 0/45// 5012 02000 OSCILLOSCOPE PROBES Switchable ) I+ 10 1262 LC 010

Used Equipment - GUARANTEED. Manuals supplied If possible.This is a VERY SMALL SAMPLE OF STOCK. SAE or Telephone for lists. Please check availability before ordering.

CARRIAGE all units t16. VAT to be added to Total of Goods and Carriage.

STEWART of READING110 WYKEHAM ROAD, READING,

AIL Telephone: (0734) 268041.BERKS RG6 1PL

Fax: (0734) 351696VISA

Callers Welcome 9am-5.30pm Monday to Friday (until 8pm Thursday)


Chelmer Valve CompanyWorldwide supplier

with 30 year's experience

Electron tubes: Transmitting,Industrial, Microwave, Audio,

Receiving, Display, etc, etc. For Maintenance, Spares or

Production. Semiconductors: Transistors,

Thyristors, Diodes, RF, Power I/C's,etc.

We have one of the largest stocksin the U.K.

* TRY US! *

FAX. PHONE, POST OR TELEX YOUR REQUIREMENTS

1 30 NEW LONDON ROAD, CHELMSFORD,ESSEX CM2 ORG, ENGLAND

Telephone: (0245) 355296/265865Telex: 995398 SEEVEE G Fax: (024S) 490064

CIRCLE NO. 124 ON REPLY 3RD

PEAK DEVIATION METER

PEAK DEVIATION EPEE

BI 0 0

"`"'"nail """t)te'"))"

The Peak Deviation Meter with its high impedanceprobe provides off -air monitoring in conjunctionwith a receiver. The unit includes a 7.5kHzdeviation standard based on a Bessel null and thusprovides a fundamental calibration standard.

010 Outlet Distribution Amplifier 4 *Advanced ActiveAerial 4kHz-30MHz *Broadcast Monitor Receiver150kHz-30MHz *PPM10 in -vision PPM and ChartRecorder *Twin PPM Rack and Box Units *Stabilizersand Fixed Shift Circuit Boards for howl reduction *StereoVariable Emphasis Limiter 3 *Stereo Disc Amplifier*Broadcast Stereo Coder *PPM5 hybrid. PPM9 micro-processor and PPM8 IEC/DIN -50/+6dB drives andmovements

SURREY ELECTRONICS LTDThe Forge, Lucks Green, Cranleigh,

Surrey GU6 7BG.Telephone: 0483 275997. Fax: 276477.

JPG ElectronicsInceitol toiohlal nainE a incl. 225V.9 10.5-0-10.5 SL952 VHF homing aniplitici LE In oo c

priniam LL260-285 heconclai c £29.95 mounting parkagc wit li data .her i I .95LEDs 3iiim or 5innE red or green 6peach AM2 7 S 0 2...f.1 .25 each 90p 111(1+Yellow I 1 p each CD40074,13 10pHigh intensity red. green or yellow imm .... 30p each 100-M6p. 1000+)(:able ties. 1 p each, £5.95 per 1000, £49.50 per Sinclair light gun terminated with a judo plug and

10,000 PP3 clip gives a signal when pointed at 50hzSmall stepping motor 4 phase 12v 7.5' step 50 flickering light with output wave form

ohms £8.95 than £3.95SAA 1027 stepping motor driver chip £4.95 DC -DC convertor Reliability model V I 2P5 12v in 5vHigh quality photo resist copper clad epoxy glass 200ma out 300v input tooutput Isolation with

boards data...£4.95 each or pack of 10 £39.50Dimensions single sided double sided Hour counter used 7 digit 2411v ac 50 Hz £1.45334 inches £0.95 £1.07 QWERTY keyboard 58 key good quality switches-I x8 inches £2.40 £2.68 £5.006,12 inches £517 Airpax A82903 -C large stepping motor 14v 7 5' step12,12 inches £10.66 27ohm Cotirnm dia body 6.3mm shaft...£8 95 orRechargeable batteries £200.00 for a box of 30A A ( H P 7 I AA 50OrnAH with solder Polyester capacitors box type 22.5rtn lead pitch

500mAH £0.99 rags £1.55 0.9uf 250vdc 18p each 14p...100+ 9p...1000+AA 700mAH £1.95 C(1 -1P1 I) 1.8A H £2.20 luf 250Vdc 20p each 15p...100+ 10p...1000+C 2 A H with solder D(HP2) 1.2 A H £2.60 2.2uf 250vdc 30p each 20p 100+ 15p...1000+

tags £3.60 P P 3 8 . 4 V 3.9uf 1 00vdc ... 30p each 20p...100 -h 15p...1000+I) 1 A H with solder 110mAH £4.95 luf 50v bipolar electrolytic amal leads 15p

£4.95 Sub C with solder each 7.5p...1000+1/2r4. with solder lags £2.50 0.22uf 250v polyester axial leads 15p each. 7.5p

tags £1.55 1/3 AA with tags (Philips 100+A A A ( H P 1 0 1 CIA') 21.95 Philips 123 series solid aluminium axial leads - 33uf

180mAH £1.75 10v & 2.2uf 40v 40p each 25p 100+Standard charger charges 4 AA cells in 5 hours or Philips 108 series 22uf 63v axial 30p

4Cs or Ds in 12-14 hours+ 1 LPN (1. 2, 3 or 4 cells each 15p...1000+may be charged at a time) £5.95 Multilayer AVX ceramic capacitors all Swoon pitch

High power charger as above but charges the Cs and 100v 1 00pf, 1 50pf. 220p(. I 0.000pf (10n) 10pDs in 5 hour, AAs, Cs and Ds must be charged in each 5p...100+ 3.5p...1000+25 or 4s . £10.05 500pf compression trimmer 60p

Special offers. please check for availability. 40 of 370vac motor start capacitor (dialectrol typeF c e I I s 3 2 d i a x F cell with solder tags containing no pcbs) £5.95 or £4950 for 10

87mm £3.95 1.2v £410 Welwyn W23 941' 120ohm 35p each 20p 100+4 2x 1 6rnm dia 680 ohm 2W metal film resistor 4p 100+ 2p 1000+

I.2v £1.4 5 Solid carbon resistors very low inductance ideal forStick of 4 17 Immx1fimm dia with red as black leads OF circuits- 27ohm 2W. 68ohm 2W 25p

4.8v £5.95 each 15p each 100+4 cell battery 94x25rnm dia (1/2 .: cells) £3.50 We have a range of 0.25w 0.5w 1 w and 2w solidComputer grade capacitors with screw terminals carbon resistors. please send SAE for list

38000uf 20v £2.50 Intelligent 4 digit alphanumeric (5x7 dot 0.145") red8 7 00 Ouf 1 Ov... £1.9 5 0 80 00uf LED display 12 pin 0.6 inch wide package Sie-

15v...£2.95 100000f 16/...£1.50 58000uf mens type DLR1414 £2.5060v £495 each £2.00..90+ Dam sheets...£1.00

7 segment common anode led display 12mm £0.45 AMD 27256-3 Eproins £2.00 each £1.25...100+LM293 IAT5.0 low drop out iiv regulator 00220 DIP switch 3PCO 12 pin (ERG SDC-3-023) 60p

package £0.85 each 40p...100+7812 and 7912 12v I A regulators £20.00 per 10)1 Disk drive boxes for 5.25 disk drive with room for aI.M337k TO3 case variable regulator £1.60(£1.111 power supply. light grey plastic.

100+) 67,268,247mm £7.95 £49.50 for 10I jaAs FEE low leakage current 58873 £12.95 each Hand held ultrasonic remote control £3.95

(£9.95 1(1+,£7.95 100+) CV2486 gas relay. 30,10mm dia with 3 wire tern -10-E192 50 P channel mosfet...£0.4 5 BC559 nals. will also work as a neon light 20p

transistor £395 each £7.50 per IOUper 100 A23120 battery for car alarms or lighters 75p

7-11105 hex inventor £10.00 per 100 Used 8748 each £50.00 per 1110Microcontroller £3.50

All products advertised are new and unused unless otherwise stated.Wide range of CMOS ITL 74HC 74F Linear Transistors kits. Rechargeable batteries, capacitors, tools etc

always in stock. Please add £1.95 towards p&p. VAT included in all prices.

JPG Electronics, 276.278 Chatsworth Road, Chesterfield S40 2BHAccessNisa (0246) 211202 Callers welcome

150

CIRCLE NO. 125 ON REPLV CARD


NEW PRODUCTS CLASSIFIED

ACTIVE

A -to -D & D -to -AconvertersSample -rate converters. SamplePortstereo, asynchronous sample -rateconverters from Analog, namelyAD1890/91, solve sample rate anddigital data interconnection problemsin computer connections and audioapplications. They are used to converta digital input sample stream at anarbitrarily clocked or changing samplerate to an output sample rate set bythe user, input clock frequency beingsensed automatically. Incompatibleequipment such as MiniDisc, CD andDCC players, HDTV and digitalspeakers will interface simply usingthese devices. AD1890 is theprofessional device, taking samplewidths up to 20 bits, while the 1891for consumer use takes up to 16 bitsat a slightly reduced functionality.Analog Devices Ltd, 0932 253320.

Discrete active devicesDual zeners. ITT announces a seriesof two-zener packages, each having acommon cathode. They are siliconplanar diodes to the E24 standardand come in surface -mounting form.DZ23 devices dissipate 300mW at25°C and cover the 2.7V -51V rangeof voltages, while the DZ89 series hasa range of 3.9V to 200V, dissipates600mW and takes maximum testcurrents of 100mA down to 5mA.Packages are SOT -23 and SOT -89A.ITT Semiconductors, 0932 336116.

SOT23 n -p -n. Of surface -mounted n -p -n transistors in Zetex's FZT range,the FZT853 handles acollector/emitter voltage of 100V andcollector current of 6A while theFZT857 copes with 300V/3.5A, with aminimum gain of 100. Saturationvoltage of the 853 is 340mV at 5A. Alldevices in the FZT range willdissipate 3W at 25°C and all take a10A peak current. Zetex plc, 061-6275105.

Linear integratedcircuitsDual audio op -amp. Exhibiting avoltage noise density of 5.2nVhili-lzand total harmonic distortion of0.004% at a gain of 1 at 1kHz with

1Vpk-pk output, Analog's SSM2135dual op -amp is meant for +5V single -supply operation in audioapplications. It will drive 240headphones directly. Its applicationsinclude balanced line driving andreceiving and sigma -delta A -to -Dbuffering and it can be used as a low-pass filter and current -to -voltageconverter at the output of an 18 -bit D -to -A converter, for example. Analogsuggests it is most suited for use withstereo codecs in computer audiosystems. Analog Devices Ltd, 0932253320.

Temperature controller. AnalogDevices's TMPO1 is a temperaturesensor that generates a voltageoutput proportional to absolutetemperature and a control signal fromone of two outputs to indicate whenthe device is above or below a settemperature range. An on -boardreference gives a stable 2.5V outputand a sensor output with atemperature coefficient of 5mV/K atan accuracy of ±2%. Wirdowcomparators provide an open -collector output to signal when high orlow thresholds are exceeded, trippoints being resistor -programmable.TMPO1 is in an 8 -pin plastic mini -dip,an 8 -pin SOIC or an 8 -lead TO -99can. Analog Devices Ltd. 0932253320.

Buffers. A range of wide -band, low -power buffers by Calogic includes theCLM4122/4222/4322 ultra -low powertypes which exhibit a 2000V/as slewrate, 180MHz bandwidth and needonly 4mW from the supply, whiledelivering 60mA peak drive current.These devices are meant to drivecoaxial and twisted -pair caoles inopen -loop application, being specifiedto drive into 500 loads.CLM4102/4202/4302 devices slew at2500V/ps at 250MHz and take only5mA from the ±3V to ±18V supply.The 4102 is said to be an improvedversion of the Elantec EL2002 buffer.Calogic Corporation, 0256 51569.

Micropower op -amps. MAX 417 andMAX418/9 dual and quad op -ampsfrom Maxim use a nove, output stageto enable them to operate at a supplycurrent of 1.2µA maximum peramplifier. With rail-to-ral outputswings and single or dual ail working,they are well suited to battery -powered equipment. Each outputsources up to 2mA and drves a1000pF load with no externalcomponents; input bias is less than0.1pA input voltage range extendingfrom the negative rail to within 1V ofthe positive rail. Unity -gain -stable,

quad MAX418 has an 8kHz GBproduct md 5V/ms slew rate, whiethe dual and quad 417 and 418 aestable at gains over 10, have a150kHz GB product and 80V/msstewing. Maxim Integrate I ProductsLtd, 0734 845255.

Voltage dropper. Semtech'selectronic zener (EZ) dropper,formerly only in TO -220, is nowavailable in SOT -23 and TO -92packages. The devices convert 5V to3.3V or 3V for mixed circuitry, passingcurrents from 0.1A to 1A No filteringis needed and, since dissipation is2W, neither is a heat sink, although ifone is used, the device handles morethan 8A. Semtech Ltd, 0592 773520.

Bifet op -amps. Up -graded families ofbifet op -amps from TI provide 10MHzbandwidth and 40V/iis slew rates. THdistortion is reduced to 0.008% andwide -band noise voltage to11.6nVNHz. Both the TLE2070 andTLE2080 are available as single, dualand quad devices. Texas Instruments,0234 223252.

Logic building blocksWindow comparator. Hysteresisprogrammable from 3mV to 2OmVand 150mW drive capability enablesHarris's CA3098 low-cost comparatorto function as a programmableSchmitt trigger, as a windowcomparator in signal processing andin automotive sensing applications,where it will switch semiconductor orinductive devices without interfacing.Hysteresis control is by means of abias variation, other characteristicssuch as input and quiescent poweralso being controllable; the sameinput is used for strobing orsquelching. Power needed is +5V to

Narrow-spectrt.m lasers.NEC claims its NDL7408Pxmultiple -quantum well lasers tohave the narrowest spectralwidth of any available, at1.3nm. In addition, the laserswork at temperatures up to85°C with no need forsemiconductor cooling. Sincethese devices exhibit lowrelative noise intensity, lowintermodulation distortion andlinear transfer characteristics,they are well suited toanalogue systems such ascable television. Conforming toCCITT requirements, the lasersare available in 0.2mW and1mW versions. The absence ofcooling reduces currentrequirement from 1A to around35mA. NEC Electronics (UK)Ltd, 0908 691133.

+16V or ±6V at a maximum of 800iiA.Harris Semiconductor (UK), 0276686886.

Fast SCSI controller. M53CF94 fastSCSI controller design by MJE iscompatible with 53CF94 devices andsupports both SCSI -1 and SCSI -2protocols. It reduces demands on theCPU by implementing common SCSIsequences from a single commandand is microcode free. The design issupplied on tape to enable it to beincorporated into a asic, being part ofthe MEJ MacroWare range ofdesigns. MEJ Electronics Ltd, 0483505895.

Mixed -signal ICsModem chipsets. High-speedmodem chipsets from AT&T offer



Please quote "Electronics World + Wireless World" when seeking further information

cellular, voice and V.32terbo facilityand include an integratedmicrocontroller and datapump withvoice compression. The HSM devicescover 9600bit/s and 14400bit/s datarates in form factors for desktop,laptop and PCMCIA use. MNP4/5/10error -correction is included. HSMchipsets now offer three choices:data/fax, data/fax/voice ordata/fax/cellular options. AT&TMicroelectronics, 0732 742999.

Serial -data interface. Maxim'sMAX562 is a three -driver, five -receiver serial -data interface thatcopes with data rates up to 250kbit/s,guaranteed slew rate being 4V/µs. Itis designed for use in notebook andpalmtop computers and meetsEIA/TIA-562; it is therefore compatiblewith RS -232 interfaces. Acceptablepower supply is 40mW from 2.7V -5.2V in normal use, all five receiversremaining active in low -power shut-down mode in which only 60pA isneeded (10µA in complete shut-down). Packaging is 28 -pin SO andSSOP, with driver outputs andreceiver inputs on the same side ofthe device. Maxim IntegratedProducts Ltd, 0734 845255.

Optical devicesOptical -fibre modules. FORCEmodel 267 optical -fibre datacommmodules transfer NRZ digital data onsingle -mode fibre at speeds up to2.5Gb/s, supporting an optical -lossbudget of 10dB. Transmitter uses a1300nm laser diode and the receivera pin -diode detector, each unit beingin an RFI/EMI-shielded enclosure.Minimum full -specification distance is10km, but up to 25km is obtainablewith care in connections and splicing.Both units are intended for direct PCBsoldering, interfacing with PicoLogic,10k or 100k ECL levels. No setting upis needed. Aerotech World Trade,0628 34555.

Power semiconductorsMicrowave power. A microwavepower transistor from Motorola, theMRF2000-5L is meant mainly for usein up to 2GHz, in large -signal outputand driver linear amplifier stages. Itdelivers 5W output for 1W input,operating from 20V supply, as aClass -A common -emitter amplifier.Motorola Inc., (USA) 602 994 6561.

HV power mosfets. A number of newpower mosfets by Motorola use apatented high -voltage powertechnique -a multiple -ring layout thatprovides field shaping towards theedge of the die, giving enhancedvoltage blocking to protect againstsurface charges that commonly occuron the junction termination at highvoltages. Motorola's 400-600V set ofmosfets have been upgraded to usethis technique and new 800V and1000V devices are now included.

Three 800V devices have draincurrents of 4A and 7A, depending onthe package, two of them having onresistances of 30 and the third 1.40.Six 1000V devices have draincurrents of 1-10A with on resistancesof 1.30-10Q. Motorola Ltd, 0296395252.

Transient suppressors. A range ofsurface -mounted transient -voltagesuppressors handling peak powers of300W, 600W and 1500W fromSemtech includes an array, theSMDA in an SO -8 package and twosingles, SMBJ and SMCJ, in DO-214AA and DO -214 AB packages.They are suitable for data, signal andsupply bus protection and come inunidirectional and bidirectional forms.Reverse stand-off is 5-24V for thearray and 5-170V for the singles;leakage current 1-100pA, 5-800pA or5-1000RA, depending on version.Semtech Ltd, 0592 773520.

PCMCIA power switch. First in aseries of PCMCIA power switchingICs by Siliconix, the Si9710CY is aPCMCIA power interface switchoffering an on resistance down to150mQ. In response to voltages atthe cmos-compatible inputs, itswitches either 3.3V or 5V to the

Ceramic resonators. Tougherand smaller than crystals,Murata's new range of ceramicresonators includes two andthree -terminal types, equivalentto series crystals and three -terminal crystal filters, the latterhaving built-in load capacitors.The filters cover the frequencyrange 3-33MHz in standardfrequencies, with custom designsavailable. Initial frequencytolerance is ±0.5% with a stabilityof ±0.3%-±0.4% withtemperature. Two -terminal unitsoperate between 1.8MHz and33.86MHz, again with customdesigns offered. Initial centre -frequency tolerance is ±5% withstability of ±0.3%-±0.4% SurtechInterconnection Ltd, 0256 51221.

supply pin in the computer's PCMCIAslot and 3V, 5V or 12V to the flashmemory program voltage pin, therebyproviding all the voltage switchingneeded for a single PCMCIA slot. Thedevice is packaged in SO -16 form.SiliconixfTEMIC Marketing, 0344485757.

PASSIVE

Connectors and cablingMemory card connectors.Interconnection and packaging formemory cards made by Elco meetPCMCIA and JEIDA standards, beingfitted with dual contact beams, first-mate/last-break contacts on theheaders and high -temperatureinsulators to cope with reflowsoldering. Elco says it can design andmake these connectors to individualspecification. Elco Europe Ltd, 0638664514.

FiltersCeramic chip resonator. Operatingat 2-20MHz and incorporating built-inload capacitors, AVX's PBRC-Bseries of ceramic chip capacitorsmeasures 7.4 by 2.6 by 2mm,ceramic casings allowirg reflowsoldering and washing. Resonantresistance to 8MHz is 330 and 1500up to 20MHz; stability withtemperature ±0.5% from 2 to 8MHzand ±0.2% from 8 to 2014Hz for-40°C to 85°C. A new catalogue oftiming devices is available on request.AVX Ltd, 0252 336868.

HardwareRFI/EMI shielding. ElectroCoatcoatings come in chrome, satinchrome, brass, nickel, copper andgold finishes on suitable plastics.

IVC's ElectroCoat 280 is a multilayercoating of nickel and copperdeposited by electroless plating forshielding to over 80dB with a 1.5pmlayer of copper and for abrasion -resistance properties. The 380version has a top layer of chrome toprovide corrosion resistance and ismeant for use on sub -miniatureconnectors. Jigs exist for manyconnectors and customers may notneed to invest in tooling. Inco VacuumCoatings, 021-511 1115.

LiteratureAmplicon Liveline. This has beenredesigned completely, nowcontaining sections on test andmeasurement, GPIB and industrialcommunications, data acquisition andcontrol, power conversion and panelinstruments. Newest productsdescribed include a low-cost neuralnetworking board for PCs, an opto-isolated digital i/o boar, a 488.2 -compatible GPIB controller for PCs,and two software packages: TRACSprocess monitoring and control andSignal Centre data acquisition andanalyse, both these being forWindows. Amplicon Liveline Ltd,(Free)0800 525335.

Farnell catalogue. 2500 new itemsare contained in the Farnell ElectronicComponents catalogue, whichincludes new sections on networkingand data communications and safety,security and warning devices,including CCTV systems. ICs fromnearly 30 manufacturers are listedand the new Psion series 3a pocketcomputer is on offer. Editions arepublished in European languages andcurrencies and one in Australasiancurrency. Farnell Electronics plc,0532 636311.

Resonators. IQD's Crystal ProductsData Book for 1984 is increased to272 pages, now being split intoleaded and surface -mounted devices.New products this year include verylow profile SM crystals for PCMCIAcards, a new range of plastic -packaged, SM clock oscillators andsome ceramic -packaged SMoscillators for high -end and PCMCIAcard use, measuring 2.3mm high by5mm by 7.5mm. Ceramic resonatorsare available in the frequency range190kHz-20MHz. IQD Ltd, 046077155.

Instrument rental. Livingston has anew mini -catalogue of instrumentationand data recording. Instruments fromRacal, Yokogawa, Sony, Fluke,Siemens and Graphtec are described,but additionally the catalogueprovides an assessment of theeconomics of renting or otherwiseacquiring data recorders, bearing inmind utilisation, obsolescence anddepreciation. Livingston Hire Ltd, 081-943 5151.



Power suppliesDC -DC converters. Three -watt DC -to -DC converters from Greshamaccept 2:1 inputs from 4.5V to 72V,which makes them suitable for usefrom 5V bus supplies, 12V and 24Vbatteries and in 48V communicationssystems. They are plasticencapsulated, are the same size asstandard 24 -pin dil packages, nutneed only convection cooling andsuffer no derating at temperaturesfrom -25°C to 71°C. There are sixsingle and dual output voltages of 5V,12V and 15V and four input -voltageratings. Gresham Power ElectronicsLtd, 0722 413060.

3.3V/5V converters. LinearTechnology's LT100 step-up DC -DCconverter is programmable for either3.3V or 5V, provides 250mA at 5Vfrom a 2V supply and takes aquiescent current of 120µA, or 10nAduring shut -down. Its power switchexhibits a saturation voltage at 1A of170mV. Since operating frequency isa minimum of 150kHz, inductors andcapacitors around the circuit may besmall, surface -mounted types.Efficiency is up to 88% at 1.8V input.The device comes in 8 -pin dip or 8 -lead SOIC packaging. Micro Call Ltd,0844 261939.

Radio communicationsproducts8GHz IC attenuator. Contained in anSOIC plastic package, Samsung'sHMP-100008-2 is a voltage -variableattenuator in a GaAs monolithic

Standard signal generator.Programmable AM/FM standardsignal generator SG -5260 byTrio Kenwood uses PLLsynthesis to produce a resolutionof 100Hz over the 10kHz-260MHz frequency range, itsoutput being from -20dBi.t to132dBp in 0.1dB steps. A digitaldisplay provides a 7 -digitfrequency readout, a 4 -digitoutput EMF figure and amplitudeor frequency modulation setting.Modulation is internal or externaland a non-volatile memorycontains up to 100 conditions tobe quickly set up. Trio KenwoodUK Ltd, 0923 816444.


microwave integrated circuit (MMIC).Using two analogue voltages of -3V toOV at less than 301.1A to controlattenuation, the device provides up to60dB over the 0-8GHz range offrequencies, without external circuitry.Units are cascadable for greaterattenuation. Operating temperature is-55°C to 85'C. Anglia MicrowavesLtd, 0277 630000.

RF power amplifier. RF2013, aGaAs heterojunction bipolar transistorby RF Micro Devices, provides up to135mW output power from a 3Vsupply and up to 800mW from 6.3V;average power output for a two-toneinput signal is 400mW from a 6.3Vsupply. Total gain, depending on theoutput matching, is 25-30dB, flatnessbeing ±3dB from 800MHz to1000MHz (±0.75dB from 800MHz to950MHz). Input voltage standing -wave ratio of less than 2:1 and betterthan -125dBm/Hz output noise powerare combined with an efficiency of47% from 6.3V and 40% from a 3Vsupply. The device is inteided for usein mobile radio and wide -areanetworks. Anglia Microwaves Ltd,0277 630000.

Mini mixer. A 200MHz-3000MHzmixer by Mini -Circuits is packaged ina ceramic surface -mount measuring0.25in by 0.31in by 0.275in. RMS-30has an IF response down to zerofrequency and can be used for up anddown conversion of RF as well as bi-phase, QPSK and l&Q modulatorsand phase detectors. The devicesconform to MIL -M-28837 standardsfor resistance to shock and vibrationand the solder pads have solder overnickel barrier for leach resistance andsolderability. Mini -Circuits Europe,0252 835094.

Transducers andsensorsHall -effect switches. Hall -effect ICscapable of operating at temperaturesup to 150°C in automotive orindustrial environments areintroduced by Allegro. Types 3121/2/3interface with bipolar or cmos logiccircuitry and incorporate a voltageregulator to handle supp'ies of 4.5-24V, a reverse battery protectiondiode, a quadratic Hall -voltagegenerator, amplifier, Schmitt triggerand open -collector output sinking upto 25mA. They are unipolar-switchingdevices for operation with bar

3 S. El- 6 O.L 'I 0 0. i324,EV Da 0

0U ri

magnets and all three are identicaexcept for differing magnetic switonpoints. Allegro Microsystems, 0932253355.

COMPUTER

Computer board levelproductsPC modules. Apex embedded PCmodules by Blue Chip are PC/ATbuilding blocks which enable amanufactirer quickly to develop etarget system while allowing him -oconcentrate on the application'srequirements. Functional supportmodules nay be added to aprocessor to form a sub -assembly ora complete system. Modules containa fully featured 80386SX or CX486slcPC/AT system with AMI BIOS,watchdog, DMA and programmablecounter/timers. There are also twoasynchronous serial ports, a mouse,keyboard and utility connectors. 512Kof flash memory is available and VGA640 by 480 resolution through asupport module. Two hard -disk andtwo floppy drives are provided. BlueChip Technology, 0244 520222.

Data acquisition board. A 68 -pinversion of National's AT -M10 -16Xhigh-performance, high -resolution,multi -function data -acquisition boardis announced. It is an analogue,digital and timing i/o board for PCsand compatible computers, having a16 -bit sampling A -to -D converter with16 analogue inputs configurable assingle -ended, pseudo -differential orfully differential inputs. National's NI -PGA instrumentation amplifier is usedin the new version to enable theboard to settle to high accuracy at100ksample/s at all gains whenscanning multiple channels. Cableassemblies are available in lengths upto 10m to connect the board to anSCSI chassis and to a wide range of

InstrumentationProgrammable PSUs. PL -Pseries power supplies fromThandar are controllable by RS-232 or GPIB, which also providereadback to the serial port of aPC without an interface card. Theseries includes single, dual andtriple models, all being made inbench -top or rack -mounted form.Main output is variable from 0 to32V, with a variable current limitto 3.1A. A 4V -6V logic output isincluded. Providing power levelsto 360W, the other new series,the TSX-P, is available in35V/10A and 18V/20A versions.The hardware and software ofthe RS -232 interface alsosupports an extended multi -instrument mode ARC, in whichup to 32 instruments may belinked and computer -controlled.Thandar Instruments, 0480412451.

Digital potentiometer. TheDigipot series of digitalpotentiometers by ControlTransducers now includes modelMD, a miniature unit that detectsdirection, position and speed.Output is a square wave at aresolution of between 100 and1024 pulses per revolution, dualchannel, with an optionalmarker/index pulse. Shaft sizesfrom 3mm to 8mm areacceptable and there are manymounting choices. The unithandles shaft play of ±0.01in andhas a screwed housing. Powerrequirement is +5V at 40mA.Control Transducers, 0234217704.




50 -pin signal -conditioning andtermination accessories. A full rangeof software is offered. NationalInstruments UK, 0635 523545.

Development andevaluation80386EX simulator. A range of 32 -bitdevelopment tools by Systems andSoftware now includes the386EX/SIM, a software simulation ofthe complete chip. It allows engineersto develop software in assembler, Cor PUM, including routines that usethe on -chip device with source -leveldebug facilities while hardware is stillbeing developed, before chips andboards become available. Code -debugging can go ahead, in manycases showing faults not detected by

LabVIEW graphics. UsingNational's new LabV!EW PictureControl Toolkit, users are able togenerate their own front -paneldisplays for the Lab VIEWcomputer -instrumentationsoftware. Possible displaysinclude bar graphs, polar plotsand Smith charts, and objectssuch as robot arms can beanimated. Images are describedby graphics instructions storedas a series of drawingcommands. There is also alibrary that implements a set offunctions that take an existingpicture and add new instructionsto create a more elaboratepicture. National Instruments UK,0635 523545.

ICE. The simulator replicates alldevices, including master and slave8529 interrupt controllers, 8524interval timers, 8250 uarts,synchronous serial units, etc, and upto 4Gbyte of memory. ComputerSolutions Ltd, 0932 829460.

87C750 development. MicroComputer Control offers the 87C750 -SDK, which is a softwaredevelopment kit for Philips's 87C750microprocessor, containing all toolsneeded to create and executeprograms. There is a multi -windowtext editor and an assembler thatfinds syntax errors, converts sourcecode to machine code and generatesprogram listings. The kit also providesa software simulator and debugger torun programs on a PC without extrahardware. An optional Micro -Ccompiler is available. LogicomCommunications Ltd, 081 756 1284.

8051 emulator. For those workingwith the 8031/51 microcontroller, theMicro AMPS in -circuit emulatoroperates up to 16MHz, has 64Mbyteof battery -backed memorypartitionable in 4K blocks betweenprogram and on -board or off -boarddata, 64K hardware breakpoints inprogram and extended data memory,single-step and software trace. MicroAMPS Ltd, 0483 268999.

In -circuit emulator. Biceps5l is anin -circuit emulator that supportsalmost all 8051 derivatives, combiningthe features of an eprom emulatorwith those of a full -function ICE. Itreplaces the eprom in a test circuitwith 64Kbyte of emulation ram that

may be partitioned in 4K blocks. Thepackage includes high-leveldebugging, hardware breakpointcapability, cross assembler and real-time trace buffer that can beinterrogated on the fly. A singleadaptor allows the unit to emulatealmost all 8051 derivatives. MicroAMPS Ltd, 0483 268999.

Eprom emulator. MicroRom fromSquarewave is a conventional epromemulator, but is contained in an11mm high module that plugs into aPC's eprom socket, eliminating theusual ribbon cable and often obviatingthe need to remove the target board.After programming, the device mayremain in circuit, being non-volatile. Itis usable with any computer fitted witha Centronics printer port and down-loads a 512Kb file in two seconds.Squarewave suggests it be regardedas an eprom with a built-inprogrammer and an unlimited numberof write cycles. SquarewaveElectronics Ltd, 081-880 9889.

SoftwareWindows data acquisition. VisualDesigner is a software package fordata acquisition designers that allowsusers to develop a PC -based systemby generating custom applicationswithout any programming. It runsunder Windows 3.1, iconsrepresenting function blocks such asi/o, graphic displays, waveformgenerators, measurement andcomparison functions, maths andlogic operations. The package istermed an application generator,holding the middle ground between

nJ Filled Waterfall Spectrum Panel

1.>d A

HorizontalHold

VerticalHold

0Amplitude 2 -angle

410Numberof Plots

STOP

4.11

rJ

L.i

fixed -function application softwareand complex programminglanguages. Custom software can be

developed simply by drawing a blockdiagram on the screen and specifyingparameters; each block thenrepresents perhaps hundreds of linesof C code. Intelligent Instrumentation,0923 896989.

Spice for RF. A new model library forRF devices that is three times the sizeof the earlier version is announced byIntusoft of California. Device modelsprovided include pin diodes,monolithic microwave ICs, H -P GaAsmesfets, Philips's BJTs and an idealcoupler. These models are usablewith any Berkeley Spice -compatiblesimulator on any computer. Thisapproach to modelling RF devices isclaimed to be superior to that usuallyadopted, in that no unrealisticparameters are needed to "force -fit"the device behaviour; custom sub -circuits account for all parasitics andmatch published s -parameters inmagnitude and phase. Any Spiceprogram is thereby enabled tosimulate linear and non-linear RFcircuits in frequency, time and DCdomains. Intusoft, 0101 (310)833-9658.

Loudspeaker design. Audiosoft ofMelbourne, Australia, has the Calsod3.00, the latest version of a softwarepackage for loudspeaker design andsystem optimisation. It simulatessound pressure and impedanceresponse of individual drivers,multiple systems and crossovers,effects of geometric layout of varioustypes of enclosure being included inthe analysis. Models of the systemsinclude loss parameters for leakage,absorption and port or passiveradiator losses. Multiple off -axisobservation points for responsecalculation can be specified, overlaidcolour screen plots being producedfor each location. A circuit optimiser isincluded, both active and passivecrossovers being optimised, andThiele -Small parameters of driversare determined from two impedancemeasurements under mass orcompliance perturbation conditions.Calsod imports data files andsupports SYSid, System One,MLSSA, IMP and LMS systems.Munro Associates, 071-379 7600.

Autorouter. ULTlroute GXR is aripup-and-retry autorouter byULTimate Technology which runsunder Windows and allows the user toinfluence the way the router performs,so customising it to his own type ofdesign. "Keep -out" areas can bespecified, both for vias and thegeneral trace. New versions ofULTIboard, a PC board designer withreal-time design -rule checks, andULTIcap are 32 -bit packages using aWindows -compatible dos extender, sothat they will run in Windows or doswith no performance loss, ULTimateTechnology Ltd, 0734 812030.

I 34 ELECTRONICS WORLD + WIRELESS WORLD February 1994

Ericsson's TBT is used in over_50 countries world-ulde

Our commitment to Training does notCommitment to training is essential in the rapidly changing con imunicationsworld. At Ericssor the importance of training has long been recognised.Since the early 1980s Ericsson has developed Technology Based Training(TBT) which allows personnel to he trained where and when the userneeds it.

Ericsson's TBT presents complex and difficult training material in aninteresting and exciting way. It is a self -paced, interactive, user-friendlyand cost effective method of enhancing technical competence.

Ericsson's generic range of IBM PC -v compatible coursewareincludes the following courses:-

Introduction to I elecommunications ISDN Overview and ISDN Advanced OSI & Data Networks and OSI Advanced X.25 and Related Protocols and X.25 Advanced C7 Signalling System Part 1 and Part 2 Understanding Modems PCM Principle Cellular Mobile Radio X.400 Overview

For further information please contact your local Ericsson office or

Marketing Department,Ericsson Systems Expertise Limited,Adelphi Centre, Upper George's Street, Dun Laoghaire, Co. Dublin, Ireland.Tel: + 353 1 2800455. After 5pm: + 353 1 2843030. Fax: + 353 1 2805914


ERICSSON

Electronic DesignsSee us

Schematic Design and Capture

1

9494 9^9

Create your schematics quickly andefficiently using EASY -PC Professional.Areas of the circuit can be highlighted onscreen and simulated automatically usingPULSAR, ANALYSER III and Z -MATCHour simulation and design programs.

Digital and Analogue SimulationModify theconfigurationand changecomponentvalues until therequiredperformance isachieved,

PCB Design

Z.4:,

".1:1

Mr±1

The design, complete with connectivity,can then be translated into the PCB. Theconnectivity and design rules can bechecked automatically to ensure that thePCB matches the schematic.

Visa, MasterCara, Amex welcome

ight First Time?on Stand 1119 CADCAM Show, NEC 8-10 March

Affordable Electronics CADEASY- PC: Entry level PCB and Schematic CAD £98.00

EASY- PC Professional: Schematic Capture andPCB CAD. Links to AINALYSER III and PULSAR.

£195.00

PULSAR: Entry level Digital Circuit Simulator- 1500 gate capacity.

£98.00

PULSAR Professional: Digital Circuit Simulator- 50,000 gate capacity.

£195.00

ANALYSER III: Entry level Linear Analogue CircuitSimulator - 130 nodes

08.00

ANALYSER Ill Professional: Linear AnalogueCircuit Simulator - 750 nodes £195.00

Z -MATCH for Windows: Smith Chart based problemsolving program for R.F. Engineers

£245.00

No penalty upgrade policy. Prices exclude P&P and VAT.

Number One Systems Ltd.Ref WW, Harding Way, St. Ives,

Huntingdon, Cambs. PE17 4WR, UK.For Full Information Please Write, Phone or Fax.

Tel: 0480 461778Fax: 0480 494042

IRCLE NO. 127 ON REPLY CARD

February 1994 ELECTRONICS WORLD+WIRELESS WORLD 15';

APPLICATIONS

Global positioning core in one ICA 11 the active circuitry needed to convert

rl12Iobal positioning information in rfspread -spectrum form to 4.309MHz final IFis contained in a single IC called theGPI010. Its manufacturer, GEC Plessey, haspublished application note AN139containing details on how to simplify thedevice's evaluation.

Input to the evaluation circuit described isC/A code signals on the GPS Ll carrier at1575.42MHz. Peripheral functions neededare a low -noise preamplifier, a 10MHzreference and a sample clock for the outputdigitiser.

Local oscillators of 1.4GHz, 140MHz and31.1MHz are provided by the GPI010 on -chip synthesizer while clocking at 40MHz isavailable for an external processor. Thereare rf and IF amplifiers with external firstand second IF filters. The third IF stageneeds no external filtering.

Output magnitude data controls the on -chip AGC loop whose time constant is setby an external capacitor. For testing, theAGC can be forced to maximum gain.

In commercial GPS applications, the firstIF filter, at 175MHz, can be implemented

using LC networking. This provides a 3dBbandwidth of approximately 20MHz. Ifsevere out -of -band interference is expected,a SAW filter may be substituted for the LCversion. The device is designed to cope withlarge interfering signals without

Intermediate frequency 0

output spectra for the GPSreceiver. Digital data isoutput in the final IF of

4.309MHz.

-10.0

-20.0

-30.0

OUTPUT dBm

-40.0

-50.0

GPS core on a chip. Thisdevice provides rf and IFamplifiers as well as the threelocal oscillators needed to

70.0convert convert the carrier 1

down to 4.309MHz final IF.

L1 VP281575.42

MHz

EXT 19LOOP 0

FILTER ,

FRONTEND

MIXER

175MHz FILTER

0 034,35 37,38

VC 0

REF '2310MHz ?-11.

LOOPFILTER

PHASEDETECTOR

1AGHz

35MHz SAW FILTER

STAGE 2AMP

MIXER

40,41 ° 43,44AGC 1 AMP

& 140MHz- .

-9

-60.0

compression. This may be important wherethe GPS receiver is co -located withinterfering signal sources such as Inmarsattransmitters.

Second 35.42MHz IF filtering is via acustom SAW device that provides a 1.8MHz

8I

CURVE I: NO SIGNAL

I

CURVE 2: -100dBm CW VP -CURVE 3: -65dBm CW VP

RESOLUTION BANDWIDTH:_

2

100KIb

I\111.1114401111,

'1114041r VP"'iffi,..,."/ 4

2 3 4

FREQUENCY MHz

5 6

AGC 2

MIXER 34.3MHzFILTER

AMP AMP

31.1MHz

DIVIDERS

SYNTHESISER

PLLLOCK

17640MHz

PROCESSOR CLOCK

321

BITE

40,41

0AGCLEV

AGC

5 71MHz

LATCH

100

110 015 13- 0- -AGC SAMPLE MAGCAP CLOCK

ADC

LATCH

SIGN

2'OF 0/P

7


APPLICATIONS

Heart of a globalpositioning receiversystem for GPS. A newapplication specific ICconverts GPS spread -spectrum code on the1575.42MHz carrier todigitised final IF.

3

S

OW9255

35MHz

S- E-

+5V

3 µ3

4

iOn

1n

NC

IF OUT

AGC LEV

+5V 3,3OSC VEE

OSC VEE

,ITT.022Cµ11'

PLL LOOP FILTER

62820 reds/s 180(= ()kHz)

R2

MAG BIT

OSC VccOSC VEE

OSC VEE

PLL FILTERAGC INTERG

C233n

0.33 AGC INTERG +

AGC FILTER

3p3

0-33nH

3p3

175MHz FILTER

1n

5,20p

00 & t t 'N N NCC CC ,. CC CC 8 L CCIW la ILI CY > Ill ILI15 x > ts )4 ,,,, ts ?..c_mitl-zzil-szitrinrinnrin

44 43 42 41 40 39 38 37 36 35 341 33

2 32 33 31

30,729

26 RF VP TR

7 27 3F VEE

RF%/EEL8

9

26 3RF Vcc 1.1n4,

10 24REF 2

12 13 14 15 16 17 18 19 20 21 22 I11 23 3

+5 47n

GP1010

NC

BITE CONTRO

33n

33n

1n

+5V

47p

Fl Vcc

LII,I uuuuuuyti : I

-,7,s > > > >LL WC

a J 2 P WCC

a

SIGN BIT

SAMP CLKPROC CLK

PLL LOCK

RF VEE

nkRF VEE

REF Vcc N T _

1177;?r-± 47n

T T

LED4110,:e.. lk

+5V

3

51

+5V

10k

I0.47p 110pC6 C7

RF EP

REF UP

+5V (VCC)

OV (GND)

bandwidth with 1dB ripple and 40dB out -of -band rejection. Tuning indicators are neededat the SAW filter output and input.

A reference source for the oscillator forthe local oscillator synthesizer can beconfigured on -chip by simply adding anexternal crystal. However the frequencyaccuracy and stability of global positioning

system receivers demands bettercharacteristics than can be provided by sucha simple source. For this reason theevaluation board is designed to accept anexternal source. This source can be derivedfrom a high-performance signal generator orfixed frequency temperature controlledcrystal oscillator.

Uses for a 2.5A transistor in SOT23y reducing saturation voltage and

U redesigning the lead frame, Zetex hasproduced a pair of SOT23 transistors thatoutperform much larger SOT223 devices.One is a 20V device with a 2.5A continuouscurrent rating, the other 50V at 2A. Gain ofthe former is at least 200 at 2A while that ofthe latter is 200 at IA. Both have a powerrating of 625mW which is over 200mWhigher than the industry norm.

At 2.5A, the FMMT618 has a saturationvoltage of typically 130mV so it is a goodchoice for switching in battery applicationssince losses will be minimal. These circuitsare from a document entitled Features andapplications of the FMMT618 andFMMT619.

The first circuit is a remote -controltransmitter. To maximise the range of thetransmitter, each photodiode needs to bepulsed at between 1 and 2A so the combined

To produce useful outputover extended distances,infra -red leds in remotecontrol units need to bepulsed at up to 2A each.High gain of the FMMT618means that no additionalinterfacing is needed andlow saturation minimiseslosses.

To minimise signal breakthrough the 5Vsupply is separately decoupled to the rf, IFand VCO stages via 3.31..tH inductors.

GEC Plessey Semiconductors, CheneyManor, Swindon, Wiltshire SN2 2QW.Tel. 0793 518000, fax 0793 518411.

load is 2 to 4A. Transmitter pulses are 16mswide. Although 4A exceeds the FMMT618'scontinuous current rating, it is within itspulsed limit of 6A.

Low saturation is also a useful parameter

in power converter applications such as theone in the second diagram. This is anunregulated isolating power supplyswitching at about 90kHz. Although tiny,the circuit has a theoretical output of IOW


APPLICATIONS

Transformer details: -Core : FX3311

W1 : 23TW2 : 23TW3 : 51W4 : 11TW5 : 11T

OV

+5V

047uF

OV

.047uF

-5V

Outputs

DC -to -DC converter with isolation. Given a smallertransformer core, this circuit could be built into a 0.3in wideDU package. Note that even though the diodes are Schottkytypes, they are the main contributors to losses.

RowDriver

Etc

Etc

Column 820Driver

820

820

Columns

3.9

FMMT618

Etc

Rows

3.9 03.9

KV8MT

FMMT618

0+5V

OV0

Part of a 32 by 32 matrix display. Column drive current is up to half anamp so transistors with good gain at high current are needed tominimise component count.

but this is limited by rectifier losses andpower dissipation capability of the substrate.Efficiency with values shown will be around85%, with most of the loss due to therectifier diodes.

In moving message displays such as theone in the third diagram, interface circuitryneeds to be as small and simple as possibledue to the large number of row and column

drivers. The circuit shown is a small portionof a 32 by 32 matrix. Each LED needs about15mA so total column current is just underhalf an amp. Because of its high gain at highcurrent, the FMMT618 needs only 5mA ofbase current to saturate to below 50mV. Asa result, it can be used for direct interfacingto low power logic registers.

Other applications outlined are printer

stepper motor driving and 1.5 to 5V powerconversion. It is also suggested that thedevices would be useful as pin drivers andFPLA programmers due to their lowsaturation voltage.

Zetex, Fields New Road , Chadderton,Oldham, Lancashire OL9 8NP. Tel. 061627 5105, fax 061 627 5467.

Inductorless voltage booster provides intermediate voltagesn barge -pump circuits are convenient for

doubling or inverting a voltage. Theyare also cheap and easy to design since theyneed no inductors. On the other hand, theydo not regulate or make it easy to provideintermediate voltages.

Adding a comparator and reference canprovide a degree of regulation and offer theability to produce intermediate voltageswithout significantly increasing complexity.As this configuration from the latest MaximEngineering Journal shows, it is possible toproduce a 3V to 5V converter that varies byonly 0.1V for loads down to 50Q.

Both reference and comparator functionsare provided within IC2. Charge pump /CI

Output parameters versus load for the regulating,variable voltage charge -pump circuit. These figuresassume a 3V supply.

Load(Q)

10k1k10050

Output(V)5.005.005.004.964.59

Ripple(mV p -p)3035100100150

has an internal oscillator whose 45kHzswitching transfers charge from Cl to C2,causing the regulated output to rise. Whenfeedback voltage at pin 3 of /C2 exceeds1.18V, the comparator turns off theoscillator via the transistor.

Because the control loop needs nohysteresis, it is set to zero but hysteresis caneasily be added via IC2. At turn on, theoscillator generates two cycles, which isalways enough to drive output voltageslightly above the desired level before feedback turns the oscillator off again.Resulting output ripple depends mainly oninput voltage and output current.

Output ripple can be reduced at theexpense of efficiency by adding a smallresistor of about Ii1 in series with C1.Ripple also depends on the value and ESRof C1. Smaller values of C1 transfer lesscharge to C2, producing smaller jumps inoutput voltage.

Maxim, 21C Horseshoe Park,Pangbourne, Reading RG8 7JW, Tel0734 845255, fax 0734 .

Charge pump circuits are normally onlygood for doubling or inverting a supplyvoltage. Adding a comparator andreference provides both regulation andintermediate output voltages.


APPLICATIONS

Fast, high-performance samplingIn a sample -and -hold system, sampling

time is limited by two consecutive events -the transition time of the multiplexer and thesettling time of the sampled signal at the output.

Application hints in the DG406/7 datasheet from Siliconix describe how toincrease the accuracy of low-level signalmeasurements by using differentialmultiplexing.

The DG406 is a 16 -channel high-performance analogue multiplexer while the407 is identical except for being configuredas a dual eight -channel device.

In a sample and hold system such as theone shown, transition time is that of themultiplexer, in this case 300ns maximum.Settling time at the load depends on severalparameters, including rbs(00 of themultiplexer, source impedance andmultiplexer and load capacitances. Chargeinjection of the multiplexer and requiredaccuracy also play a role.

Settling time for the multiplexer alone canbe derived from the model shown.Assuming a low impedance source, such aspresented by an operational or bufferamplifier, settling time of the RC networkfor a given accuracy is or.

% accuracy # of bits0.25 8 60.012 12 9

0.0017 15 11

Maximum sampling frequency of themultiplexer is,

1

fs=)41 -Settling + tTrans)

where .v is the number of channels to scanand,

= "/ = n x

lbSensor 1

1

11,1

Sensor 8 I

404.Analog

Multiplexer

DG407

Analog

InstAmp Sat

12 -BitA/D

Converter

High-performance analogue multiplexerforms the heart of a fast sample -and -holdsystem for low-level signals. Using differentialmultiplexing helps reject common -mode

Controller

rDS(on)

Simplified model of a single -channelmultiplexer. With this model, settling timefor the input multiplexer of the sample -and -hold circuit can be derived.

Given a DG406 at room temperature, for12 bit accuracy using maximum limits,

1

16(9x100S2x100x10 12F)+3(X)x10-1-s

orfs=694kHz.From the sampling theorem, to properly

recover the original signal, the samplingfrequency should be more than twice themaximum frequency of the original signal.This assumes perfect band limiting. In a realapplication, sampling at three or four timesthe frequency of the filter cut-off frequencyis advisable so,

Ultra -fast op -amp with clamplaimed to be the fastest monolithicamplifier available, the HFAI 130

from Harris has a -3dB bandwidth of850MHz and a slew rate of 2300V4ts.Distortion is low, at -56dBc, and thedevice has user programmable outputclamping to protect later stages fromdamage or input saturation.

Acconthig to the device data sheet, thereis an evaluation board for the cevicefollowing this circuit. Discussed within thesheet are optium feedback resistance,clamp cperation, PCB design and recoveryfrom overdrive. Comprehensiveperformance figures are also presented.

Overdrive recovery of the device is

f = -1 x f =173kH:

This shows that the DG406 can samplesixteen different signals with a maximumcomponent to 173kHz. Using two channelsto double sample the incoming signal alsodoubles cut-off frequency.

In the block diagram is a typical data -acquisition front end suitable for low-levelanalogue signals. Differential multiplexingof small signals is preferred since it helps toreject common -mode noise. This isespecially important when sensors arelocated at a distance.

A low rbsont, low leakage multiplexerhelps to reduce measurement errors. TheDG407 has lower than usual powerdissipation so on -chip thermal gradients arereduced. These gradients can cause errorsdue to temperature mismatch along parasiticthermocouple paths.

Siliconix, East Hampstead Road,Bracknell, Berkshire RG12 1LX. Tel.0344 485757, fax 0344 427371.

typically less than a nanosecond whilesettling time is 11 ns. Gain flatness at100MHz is 0.14dB.

Haris Semiconductor, Riverside Way,Camberley, Surrey GU15 3YQ,Tel. 0276 686886.

Ons 25ns

INPUT220MHzSIGNAL

OUTPUT

(A5 2)HFA1130OP AMP

10µF

+5V

OUT

VLGNC

Caimed to be the fasted monolithic op -amp, this device has an8.50MHz bandWdth and 2300V/µs slew rate.


BROADCAST

delivery, delay or debacle?

Digital AudioBroadcasting is seen by

many as the logicalprogression in soundtransmission for the

next century. Itundoubtedly represents

magnificent technicalprogress. However, the

packing of channelswhich is part of thesystem reduces the

independence ofindividual station

operators. It deliversthe quality but will itdeliver the audience?

Norman McLeodreports.

The politics of DABI overheard a delegate at the Radio Academy'stechnical conference last November mutter"Will DAB be another RDS...'?" RDS - theRadio Data System - was launched in a blazeof glory six years ago, as nothing less than"the greatest improvement in radio sets sincethe invention of the transistor" and "the key tothe future of radio listening". Few wouldmake such hyperbolic claims for this decid-edly minor miracle today.

DAB - Digital Audio Broadcasting - firstcame to public attention amid similarly glow-ing praise for its properties of high soundquality and immunity to interference. It hasbeen popularly reported as a spectacular tech-nology under test and demonstration condi-tions, but without much anticipation of itswider implications and real -life practicality asa consumer product.

Eureka DAB - the European version of dig-ital audio broadcasting - is an impressiveattempt at tackling the technical shortcomingsof present-day FM radio broadcasts.

The fall of FMFM has not lived up to the ambitionsexpressed by early pioneers for an interfer-ence -free high quality service intended even-

tually to replace AM broadcasting on long andmedium wave. In the fifties, when VHFbroadcasting first started, excellent monoresults could be obtained from fixed VHF/FMreceivers connected to an outdoor rooftop aeri-al, as was the accepted norm for televisionreception.

What was not foreseen at the time was theadvent of portable and mobile VHF receiverswith modest, street -level aerials. And stereowas in its infancy - the Zenith -GE multiplexsystem unheard of. Both these factors heavilycompromise FM reception quality today.

Any prospective broadcasting service mustwork well in portable and car radios. The sig-nals picked up by these receivers, alreadymuch lower than the rooftop signals for whichVHF broadcasting was originally planned,fluctuate randomly by up to 20dB as thereceiver is moved. This causes swishes andpops of noise whenever the signal falls toolow for reasonable quieting.

FM broadcasts are also affected by multi -path distortion caused by time -delayed com-ponents, due to signal reflections from build-ings and other obstacles, which arrive at theantenna varying in phase and amplitude withrespect to the direct signal'-. Multipath distor-tion of the FM signal at the antenna translates

Multipath consequences for FM:This graph shows how susceptibleFM stereo reception is to delayedreflections. The required direct tounwanted (D/U) signal ratio isshown for various time delays andresultant distortion. Seriousproblems start with delays of lessthan 511s (1.5km extra pathlength). Eureka DAB, on the otherhand, is completely resistant tomultipath content - indeedbenefits from it - until the signaldelay reaches 300µs.Source: "Analysis of MultipathDistortion in FM SoundBroadcasting", NHK TechnicalResearch Laboratories, June 1979,quoted by Hawker2.

17j

00

CC

40

30

20

10

Modulation: left signal, 15 kHz, 100%

... 1'

3'

21

-k

wAr..........._____....x... O'''^<-- -. -.)"."' "" --)*

, A11

* -- "" ...

if-lb.-, Afihe

>e-..4('' Max. distortion

0 5 10 15 20

Delay time (µs)

25 30


BROADCAST

into unpleasant nonlinear distortion of thereceived audio. When the Zenith -GE pilottone stereo system is in operation, matters getconsiderably worse:

Stereo operation worsens the signal-to-noiseratio of the transmission by 20d13;

Multiplexing makes co- and adjacent chan-nel interference more troublesome - some 15to 20 dB more protection is required when thestereo decoder is operational;

Separation between the channels is incom-plete and highly vulnerable to small errors onthe multiplex anywhere in the transmissionchain from coder to decoder.

Multipath reflections can distort the super-sonic 'S' signal to the extent that receptionwhich is tolerable in mono can be unlistenablein stereo, even when the signal level is suchthat noise is not a problem.

The Eureka DAB alternative positivelythrives on multipath effects, has no penalty forstereo working, and offers the flexibility to

trade off ouantity against quality when itcomes to allocating bits to services. It per-forms magnificently.

But it is being called into question, not byengineers. hut by the people who will have towork with it and foot the bill for the new tech-nology.

Standard struggleWhile DAB shines technically, it is clear thatthere is a political imperative to heat US tech-nical proposals: IBOC (in -band on -channel) iscurrently being researched by USA Digital,AT&T Bell Communications and AmatiCommunications Corp. IBOC systems piggy-back a digital carrier on top of existing ana-logue FM (or even AM) transmissions. Theydo not require a new frequency band, andplace firm control of the new medium in thehands of existing operators with their owntransmitters, sites and frequencies.

Although US work is at a much earlier stage

of development than Eureka. IBOC has suchstrong attractions for established broadcastersthat if they do prove viable. Eureka willalmost certainly be eclipsed. So there is a rushbordering on desperation to have Eurekaaccepted and up and running before IBOC hasthe chance to prove itself one way or theother. Adoption of Eureka inevitably meanscentralised control of a broadcasting channelfor its interleaved services. The technicalaspects of this system make it mandatory tosupply a multiplex of services from centrally -assembled data streams. The 'frequency diver-sity' properties which allow the multiplex toresist drop -outs in the incoming signal. whichare typically between 30 and 300kHz wide.make it necessary to occupy a wide bandwidthwith an interleaved signal.

Frequencies must be planned on the basis of1.75MHz wide chunks, as opposed to seg-ments an eighth of that size on FM. Thismakes Eureka planning a much less flexible

COFDM: transmitting the digitsCOFDM (coded orthogonal frequencydivision multiplex) describes the processused by Eureka DAB to send data over theradio transmission path14. The essentialfeature of this system is that a largenumber of closely -spaced carriers areused, each of which carries a relativelyslow data rate. In Mode 1 DAB, asdemonstrated by the BBC on 226MHz, atotal of 1536 carriers are used 1kHz apart .

The total RF bandwidth occupied istherefore 1.537 MHz, but a further 250kHz guard band is needed betweenadjacent blocks, as the DAB signal doesnot cut-off very cleanly at the edges of thespectrum, and is prone to spread due tointermodulation effects. It has been notedthat TV transmitters carrying DAB may bebacked off some 9dB on pep output forgood IM performance.

Two bit QPSK symbols are transmittedon each carrier at intervals of 1.246ms,made up of 1ms active symbol duration,and a 246µs guard interval period. The bitrate per carrier is therefore just over 1600bits per second, making the total availabledata rate some 2.46Mbits/s. Of this,1.09Mbits/s was used in the BBCdemonstration to transmit audio from thefive national radio networks at threedifferent quality grades - the rest isaccounted for by error correction andidentification data.

Each slow COFDM carrier is unruffledby delayed signals arriving up to 1.2 timesthe guard interval later than the directsignal - in this case 300µs, correspondingto an extra path length for the delayedsignal of 90km. Indeed delayed signalscaused by multipath reflections add to,rather than detract from, the receivedsignal quality. The 300µs permissibledelay is well beyond the delay

encountered in multipath conditions froma single transmitter, and opens up thepossibility of Single Frequency Networks(SFNs). Transmitters carrying the sameprogramme selection can all operate on thesame frequency, and provided thedifference in path length between twotransmitters does not exceed 90km, andthe transmitted signals are synchronisedwith each other, the effect on the receiverof multiple signals is wholly beneficial.

COFDM in Mode 1 is limited toapplications below about 350MHz becauseit is sensitive to frequency shifts caused bythe Doppler effect. A shift of 1/20 of thecarrier spacing (50Hz) increases therequired carrier -to -noise ratio by 4dB. It isconvenient to remember that at 300MHzthe Doppler shift in hertz has the samenumeric value as the speed of a moving

receiver in m/s. So signals picked up in aspeeding car or a fast train moving at50m/s (about 112mile/h) directly awayfrom or toward the transmitter will suffer a50Hz shift at 300MHz, and pro rata athigher or lower frequencies. At the226MHz frequency proposed as a 'parkingband' for DAB, the 'velocity loss' will notexceed 4dB until the vehicle is travellingat nearly 150mile/h.

Modes 2 and 3 define carriers 4 and8kHz apart, and can be used withoutunacceptable Doppler limitations up toabout 1.4 and 2.8GHz respectively. Mode2 may be applicable to local broadcastingand limited -area single -frequencynetworks, and allows a guard interval of62.5µs. Mode 3 suits satellite or hybridsystems in the `1,' and 'S' bands - theguard interval here is 31.25µs.

COFDM In practice: This shows the COFDM (coded orthogonal frequency division multiplex)system working in Mark 1 (FDMA) mode. Each programme permanently occupies one carrier inevery six. FDMA was used in the early Rennes and Geneva DAB demonstrations.The Mark 2 TDMA mode - used in the BBC and NAB DAB demonstrations - uses all the carriersinterleaved in the time domain for all the programmes, and seems to be more resistant toimpulsive interference. But it would preclude radiating a set of skeleton carriers in an attempt tointroduce more flexibility for individual channel operators.


BROADCAST

process when it comes to matching local radioservices to particular communities. It meansfor instance that a 'one -for -one' guarantee -that every current FM service can have a com-parable DAB outlet - would be impossible toachieve economically. The shape and structureof the radio market will change fundamental-ly with Eureka DAB.

Private fearsEureka means bundled broadcast channels.Independent operators running just one or twoservices do not want another four or five start-ing up in parallel with them, while stand-aloneservices in rural areas or small markets do notwant to have to pay for a six -service installa-tion when all they require is one, alwaysassuming that a frequency band would beavailable for them. Eureka DAB can run sin-

gle -frequency networks with great ease overwide areas, but the benefits of this sort ofworking turn to dust when a different pro-gramme is required in a different area, and theprotection ratios required end up much thesame as FM3.

A recent report from Germany, the birth-place of DAB, began with Ursula AdeIt of thePrivate Broadcasters' Association sendingshock waves through the recent IFA show inBerlin by declaring that "for the private radiooperators, Eureka 147 is dead4.- AlreadyARD, the German state broadcaster, hascaused gloom in the Eureka lobby byannouncing that it could not afford to imple-ment DAB until 1997.

Now private broadcasters are beginning tofret about who is going to pay for it, whetherlisteners really want it, or whether it might

Musicam: cutting the bit budgetA decade of work on bit rate reduction hasmade the compact disc, shifting 1.4Mbit/sfor two channels of audio, look highlyextravagant. Television's nicam systemmanages with 728 kbits/s using a simplescaling technique, but Musicam - in theform known by the snappy title "ISO-MPEG Layer 2: IEC CD 11172-3" offersbetween 192-256kbits/s per stereo channelfor quality closely approaching that of theoriginal source.

The block diagram of the audio encoderis shown below. The heart of the process isthe `psychoacoustic model' -a devicewhich exploits the fact that the thresholdof hearing is programme -dependent. In the

presence of a 500Hz tone at 70dBA, forexample, the threshold of hearing is raised10dB between 300Hz and 1kHz, whilebetween 400 and 600Hz energy needs tobe above 40dBA to be detected.Ultimately, 'perceptual coders' such asthis one aim to produce a noise floorwhich pumps up and down in 32 sub -bands in such a way that it is alwaysmasked by the programme material aboveitI2,13.

The bit allocation for each sub -band isdetermined by a calculation involvinginput from a 1024 -point FFT scan capableof detecting the difference between sinu-soidal and noise -like energy and adjusting

PCMaudio

samplesCodedaudio

Filterbank quantizer frame48 kHz bitstream

32 subbands andcoding packing

psychoacousticmodel Bit allocation ISO 11172-3

Layer II

(a)

kFormatter

Simplified blockdiagram of basicDAB audiocoder

Reformatterframe

unpacking econstructioninverse

filterbank32 subbands

DAB coded ISO 11172-3 PCMBaseband Audio Bitstream Audio Samples

Signal ISO 11172-3 48 kHz

Layer II

(b)

Simplified block diagram of basic DAB decoderSource: see Ref. 13.

share the fate of Digital Satellite Radio, withjust 100,000 listeners after five years. Reportsfrom Washington also put the very existenceof a US digital radio business in doubt giventhe current business environments.

The development of DABThe Eureka Project EU I47 on digital broad-casting was launched at the EuropeanConference of Ministers in Stockholm in1986, initially for a four-year plan of researchand development between 1987 and 1991.Late in 1991, the second phase began. It hasbeen looking into system specifications, appli-cation -specific integrated circuits and detailsof mass communication services additional tobroadcasting which could be provided. Thesemight include public or subscription servicescarrying data, for which there is some spare

masking thresholds to suit. The data fromthe sub -band filter is re-quantised to main-tain just enough resolution (maybe onlytwo or three bits) for the quantisation noiseto be inaudible below the masking thresh-old for each band.The perceived sound quality of this verycomplex process can be traded against bitrate to a fine degree. The BBC demonstra-tions offered the following options:

256kbit/s independent stereo for Radio 3;224kbit/s joint stereo for Radios 1 & 2;192kbit/s joint stereo for Radios 4 & 5.

The 'joint stereo' process exploits theredundancy between left and right signalsof a stereo pair to provide better qualityfor a given bit budget than could beobtained with two independent monochannels each running at half the rate. Theoverall bit budgets that can be assignedvary from 32kbit/s to 192kbit/s for a singlechannel, the latter allowing headroom forfurther processing after reception, or from64kbit/s to 320kbit/s for a stereo channelor pair of services.

It is very difficult to assess the perfor-mance of Musicam at the bit -rates offeredwhen the original, uncompressed signal isnot available for comparison, and there iscurrently such a shortage of DABreceivers that it was not possible to do anA/B listening test inside BroadcastingHouse.

Any lack of transparency has a funda-mentally different quality from familiaranalogue shortcomings, although listenerswho have taken part in demonstrations saythat once the ear has learnt what to listenfor, deficiencies can be more readilydetected. In the BBC coach, any sonic dif-ference between DAB and undegraded FMwas very subtle at the bit rates chosen fordemonstration.


BROADCAST

Demonstrating DAB"The digits deliver the goods - they solveeverything," said one engineer at the BBCdemonstration of Digital AudioBroadcasting in central London in earlyDecember. From a technical point of view,the coach ride - with DAB or FMselectable on headphones - was impres-sive, but with reservations.

We were invited to compare FM recep-tion from Wrotham, 20 miles to the East inKent, with DAB reception from muchcloser transmitters atCrystal Palace andAlexandra Palace runningsimultaneously. DABreception was solid all theway, while FM was subjectto the familiar noise andfading which mars muchcity reception. But we werenot, it has to be said, com-paring like with like.

I put this to the BBC,suggesting that a fairerdemonstration would havebeen to compare DAB withFM reception from CrystalPalace (the BBC's GLRtransmitter on 94.9 MHz),or alternatively to radiateDAB only from Wrothamso that it suffered the samedegradation as the FM net-work signal. I was told thatthe BBC wanted to showDAB on the network ser-vices with their variousprogramme content, forwhich there is no centralLondon transmitter, andthat to bring DAB into cen-tral London from Wrothamwould require a transmitterpower of at least 10kW,which was not at theBBC's disposal.

Originally, the tests werewired to provide unpro-cessed, uncompressedfeeds to the DAB transmit-ter while leaving theÒptimod' multi -band sig-nal processors in circuit onthe FM system. Thisproved to be at unwisemove: listeners commentedon differences in the soundquality of the two services,attributing them (wrongly)to inherent characteristicsof FM versus DAB, where-

as in fact they were observing the effectsof the Optimod units, which make pro-gramme material seem denser, louder andsuperficially more impressive. So DABgot the same processed feeds as FM afterthe first day.

This meant that, interference apart, DABaudio quality sounded practically identicalto FM. It may indeed have been sold shortby having been fed first from a Nicam linksampling at 32kHz and therefore limiting

the audio bandwidth on both systems to15kHz. To get digital quality above, andnot just on a par with, FM may well needsome re -engineering of distribution net-works in the future. And we have beenreminded that most people prefer limiteddynamic range on their radio material,with the exception of a few readers of Hi-Fi News' I, so that CD levels of dynamicrange are positively unwelcome to mostconsumers.

Broadcasting House

Radio 1

Radio 2

OPTIMODprocessor

Radio 3

OPTIMODprocessor

Wrotham Transmitting Station

Band IIFMmodulator power amplifier

stereo b

Band IIFM signals

FM Band IImodulator power amplifier

NICAMlinks FM Band II Radio 1

Radio 2modulator power amplifierRadio 3

FM Band II Radio 4

modulator power amplifier

Radio 4

OPTIMODprocessor

Radio 5

OPTIMODprocessor

OPTIMODprocessor

monoPCMlinks

Brookmars Park Transmitting Station

AMmodulator

Medium Wavepcwer amplifier

mb

TMedium WaveAM signal

Radio 5

MUSICAM Hsource coder

MUSICAMsource coder

stereocable links

MUSICAMsource coder

MUSICAMsource coder

MUSICAMsource coder

p

e

e

Crystal Palace Transmitting Station

COFDMchannel coder

RFmodulator

inputfilter

frequencyconverter

Band IIIpower amplifier

Band Di- outputpqweramplitier filter

226 MHzDAB signal

outputfilter

Band IVdistribution

signal

Alexandra Palace Transmitting Station

ow -noiseamplifier

frequencvconverte-

Band IIIpower amplifier

outputfilter

226 MHzDAB signal

Arrangements for the London DAB demonstration. This shows the arrangements made to distribute anddemonstrate DAB by the BBC in central London, with two important omissions. The input to theMusicam coders was transferred to the output of the 'Optimod' processors to avoid misleading listenershearing the effects of the processor. And the process of distributing audio to the DAB network washandled by a Nicam system operating at 32kHz sampling rate, which is not shown on the diagram.


BROADCAST

bipolarf, = 5 GHz

RF/IF

RF MixerIFLO

cmos/bicmos cmos/bicmosclock speed = 50 MHz clock speed = 40 MHzgates = 50K gates = 100K

DSP I

A/DFFT

Synch.

cmosclock speed = 40 MHzsemi custom audio DSP

DSP II

De -interleaveViterbi Dec.OPSK Det

SourceDecode

Proposed four -chip DAB receiver. This shows a possible architecture for a practical DABreceiver. RF functions are handled by high speed bipolar technology using an oxide or trenchisolated process. The first DSP incorporates the A/D converter, FFT algorithm andsynchronisation circuitry. The next two ICs take in the Viterbi decoder, de-interleaver, QPSKdetector and source decoder. The hope is that source decoders will become off -the -shelfdevices if the DAB Musicam audio compression standard is widely implemented elsewhere.Source: Ref. 6

capacity. The project is a consortium of part-ners including broadcasters, PTT administra-tions and consumer electronics interests.

In the UK, Michael Heseltine announced thesetting up of a DAB Forum in the Commonson 16 February 1983, declaring that he wasanxious to ensure the benefits and opportuni-ties of DAB were made available in the UK.Broadcasters, equipment manufacturers andother interested parties were invited to join. Ata meeting with opposite numbers representingFrance, Germany and the Netherlands, theyconsidered factors necessary for the successfulearly introduction of DAB. They concludedthat success depended on the following:

an early move to a single frequency bandwith sufficient spectrum for all public and pri-vate broadcasting systems; new programmes on DAB, not just simul-casts of current channels;additional services, such as data broadcasting; new ways of financing digital broadcasting,including encryption and pay -per -listen;the support of car manufacturers to create apool of DAB car radios; a legal framework which provides incen-tives and allows for innovation; conditionswhich will provide sufficient rewards forDAB pioneers.

Alongside these committee recommendationsremains the fundamental question: how muchwill DAB cost to transmit?

At the transmission end of the chain, MikeThorne, of UK transmission providers NTL,the privatised wing of the old IBA, has beenbold enough to put a figure on transmittingDAB to a large urban area, although he wouldbe the first to admit that these are projectedcosts at this stage, and not a bid for a contract.

NTL's costs assume that an existing trans-mitter site can be used, with space availableon the mast for the 230 MHz antenna system.and room below in the building for the appa-ratus. Given this, construction costs for a sin-gle 1kW station are put at £450,000, and therunning costs - use of mast, rent of building,electricity, BT lines and maintenance etc. -would amount to £70,000 pa. To provide a

`total broadcast contract', including payingback the capital costs over some 8 - 10 years,and a sixth of the running costs above, wouldcost each of six operators £25,000 pa.

To put together five 250W transmitters in asingle frequency network (SFN) would costeach operator about double the amount for asingle transmitter (£40 - 50,000 pa), butwould provide much more solid coverage andbetter frequency use. This is not dissimilar tothe present-day costs of a comparable FM ser-vice from the same company.

Receiver costsAs for the receiver, the current state of devel-opment is known as the third generation, andconsists of a substantial rack of equipmentdrawing 2.5A from a 12V supply, and costingover £3000. A four -chip future receiver sellingfor the price of a good CD player (£150 -£200) has been sketched out but is still a longway off6.

More immediately, an assessment of likelyparameters for a start-up phase in 19957 antic-ipates that first generation consumer DABproducts will contain around ten specific inte-grated circuits together with their peripheralcomponents.

The first sets will be expensive, on a par,perhaps, with the early fax machines ormobile phones. It is highly unlikely that a £50DAB 'Walkman' radio will be a practicalproposition for a few years. DAB and FM willnot compete initially for mass -appealreceivers.

Estimates of demand and penetration levelshave been produced9. The most hopeful pro-jection does not foresee DAB reaching 50% ofthe marketplace before 2010. This is threeyears later than the date when some are rec-ommending that FM stations are closed downto make way for its.

DAB and the radio businessIn the long term, Eureka DAB, with its cen-tralised transmission system, will profoundlyaffect the nature of the marketplace in radiofor both transmission and programme services,favouring national networks and large -area

broadcasters over smaller companies. ThatDAB is seen not as an addition to FM but inthe long term as a replacement for it implies apossible threat to FM stations.

If an existing operator merely re -broadcastscurrent programmes on DAB, he possiblyfaces the additional costs of transmitting thenew service, but without the audience and rev-enue to fund it. There may however be someprofit in subsidiary data services. Previousattempts to market Radiotext data services ontop of FM broadcasts have been unsuccessful.

On the other hand, if an established or newoperator decides to offer a separate audio ser-vice on DAB in the hope of attracting new lis-teners and sources of revenue, he will have tobear the costs of producing this alternativeprogramme. This is in addition to transmittingit while the service builds up to profitable pro-portions.

Future developmentsThese possible business shortcomings in theDAB system have focussed thoughts on mod-ifications. Dr. Brian Evans" has suggestedthat were the Mark 1 FDMA interleaving pro-cess to be adopted, it ought to be possible toerase five of the six sets of carriers from themix in order to permit individual transmittersto provide individual services without the needfor inter -broadcaster synchronisation ofsources.

There are penalties with this system in that itis no longer possible to reach the theoreticalminimum spacing between carriers, and theFDMA system is generally less efficient thana full TDMA multiplex at resisting interfer-ence. But it does address the handicap of cen-tralised transmission without losing the keyquality features of DAB.

Improvements to DAB's non -gracefuldegradation are under examination. To have asystem fail completely, even if only 1% oflocations are affected, is still a serious handi-cap compared to the more graceful degrada-tion of analogue systems. AM carriers or someanalogue 'helper' signals which could beadded to the DAB multiplex to save the daywithout themselves taking up too much band-width or power might smooth out this techni-cal rough edge to everyone's satisfaction.

DAB spectrum use and frequencyplanningPlans for the UK interim DAB service onBand III propose two I .75MHz-wide blocksfor national services, one set for the BBC andone for the commercial sector, and five furtherblocks for local and regional coverage. Amaximum of 42 services could be potentiallyavailable if all blocks could be received in agiven location.

In terms of programme packing density permegahertz, DAB is very broadly equivalent toFM when set against the local radio sectionsof the FM band (94.6 - 97.6 and 102 -105MHz). Typically six DAB services areprovided in 1.75MHz of spectrum. WhereBand II is well -used, it is also possible to pickout six or so FM services across I .75MHz on


BROADCAST

a reasonably good tuner and aerial. The noisebandwidth per service used for DAB s/n cal-culations is typically 300kHz, slightly greaterthan the bandwidth of an FM transmission(240kHz).

National networksFor national networks, DAB looks splendidlyefficient at frequency use, since the sameI .75MHz slab of spectrum can be re -used adinfinitum across the UK to provide five or sixnational services. The current five national FMservices consume 11MHz between them -more than half of the Band II allocation.

The ability to extend national networks overan SFN stops, of course, at national bound-aries. Maps like the one shown have beenproduced showing Europe painted accordingto the four-colour map theory. This wouldenable four blocks to be used for nationalSFNs with only minor problems in places likeLuxembourg or Liechtenstein. National blocksassigned to other countries could, of course, bere -used by smaller regional or local servicesaway from the frontiers.

DAB fails - abruptly and completely -when the incoming carrier -to -noise ratio fallsbelow 10 - 14dB. Above that, there is solidreception independent of signal strength. FMsystems degrade gracefully as the signal fails,so that for planning purposes a median curve(signal exceeded in 50% of locations, 50% ofthe time) can give a good rule of thumb forsetting 'coverage'. Anyone receiving less thana fair share of signal may get degraded recep-tion, but will still be able to follow the service.

Because DAB does not behave in this man-ner, calculations for DAB planning have to bebased on generous margins - up to 99% prob-ability - on top of median values which lookimpressive in isolation. Calculations haveshown that assuming a receiver noise figure of8dB operating at 200MHz from a practical carantenna, the required DAB field strength mustbe at least 22dElliVim in the vicinity of theaerial, which may be barely 2m off theground15. The familiar planning curves forFM broadcasting assume the use of an anten-na 10m highly. A correction for this discrep-ancy is open to debate but we will assume12dB, bringing the signal requirement at 10mto a modest 34dE4tV/m (501.1V/m). On a fixedFM tuner with an outdoor aerial this level ofstrength will produce tolerably quiet monoreception, but mobile FM reception would bechoppy even with a good receiver.

Onto this figure must be added a factor toincrease the probability of the signal reachingthis minimum value from 50% to, say, 99%.This would amount to I 9dB, bringing therequired median level to 53dBµV/m, verysimilar to the field strength required for stereoreception of the FM service. However, wherean SFN is in operation (and this is the mostlikely situation) there is a factor working toimprove matters because the receiver aerial isbeing illuminated by signals from more thanone direction, greatly increasing the chancesof successful reception. An advantage of 6 -10dB might be expected from this source. But

Europeanfrequency reuse:just four DABfrequency blockscould provide anational network ofprogrammes -upto six per block -in each countrywithout co -channelinterference.

against this, propagation curves for Band II(CCIR Rec 370) need to be scaled down by6dB when applied to 230MHz predictions;this is a function of the frequency, of course,and not the system, but it works against DABin Band III (though would work in its favouron Band I).

Man-made interference in populated areas isbecoming significant. This can increase thenoise floor of the received signal by 7dB ormore. The effect is worst in dense urban areas,but generally signals are planned to be

stronger in these locations as a matter ofcourse. Nevertheless, a value of 4dB has beensuggested as an overall margin to allow forcurrent levels of pollution from computer anddigital electronic equipment.

In general, it seems that a DAB SFN of per-haps five 250W DAB transmitters could pro-vide substantially better coverage over a largeurban area than a single FM transmitter run-ning 2-5kW. Were the DAB transmitters to beco -sited with existing UHF television trans-mitters where aerial space above and accom-modation below are already available, initialprojections show that DAB may offer slightsavings over a comparable number of FM ser-vices. But there are still trade-offs and marginsin DAB planning.

References1. RDS - an ear to the future, BBC EngineeringTraining Department, 19882. Pat Hawker, How serious is multipathdistortion?, Wireless World, April 1980, p.453. Gerd Petke, Institut fur RundfunktechnikGmBh, Planning aspects and plans forterrestrial DAB implementation, Proc. ofMontreux DAB Symposium, 1992, p.167(EBU)

4. Fundamental Questions Hamper GermanDAB, Radio World, 23 Nov 1993, p.245. USA DAB Confronts New Obstacles, RadioWorld, 23 Nov 1993, p.206. McComas, Kady & Warren, DelcoElectronics Corp., Manufacturing DABautomobile receivers, Proc. of Montreux DABSymposium, 1992, p.235 (EBU)7. Manfred Halbe, Philips ConsumerElectronics International by, DAB receivertechnology and implementationconsiderations, Proc. of Montreux DABsymposium, 1992, p.121 (EBU)8. FM Stations may close in spectrum shake-up, Electronics World and Wireless World,September 1993, p.7089. M. Carrasco & A. Lange, Idate consortium,What markets for DAB in Western Europe,Proc. of Montreux DAB Symposium, 1992,p.245 (EBU)10. Dr. Brian Evans, Tantara Tek Ltd., Ensuringa bright future for independent DAB,September 199311. Reg Williamson, The Limiting Factor, Hi-FiNews & Record Review, October 1993, p.7312. Fred Wylie, APT, Digital Audio DataCompression, Audio Media, September 1993,p.8413. Gerhard Stoll, Institut for Rundfunktechnik,Munich, Source Coding for DAB and theevaluation of its performance: a majorapplication of the new ISO Audio CodingStandard, Proc. of Montreux DAB Symposium,1992, p.83 (EBU)14. B. Le Floch, CCETT, Channel coding andmodulation for DAB, proc. of Montreux DABSymposium, 1992, p.99 (EBU)15. C.P.Bell & W.F.Williams, BBC ResearchDept., DAB: Digital Audio Broadcastingcoverage aspects of a single -frequencynetwork IBC, July 199216. Recommendations and reports of theCCIR, Volume V, Rec. 370-5, ITU, 1990


Surplus alwayswanted for cash' THE ORIGINAL SURPLUS WONDERLAND! Surplus always

wanted for cash!

LOW COST PC SPECIALISTS - ALL EXPANDABLE - ALL PC COMPATIBLE8088 XT - PC99

256k RAM - expandableto 640k

4.7 Mhz speed360k 5-1/4" floppy2 serial & 1 parallel portsMS-DOS 4.01

Factory burnt -in Standard 84 key

keyboard 12" green screen

included In good used condition

Optional FITTED extras: 640K RAM £39. 12' CGA colourmonitor with card £39. 2nd 5-1/4' 360K floppy £29.95. 20mbyte MFM hard drive £99.

offiYE99.00,,FLOPPY DISK DRIVES

51/4 " from £22.95 - 31/2" from £21.95!Massive purchases of standard 51/4' and 31/2' drives enables usto present prime product at industry beating low prices! All units(unless stated) are removed from often brand new equipmentand are fully tested, aligned and shipped to you with a 90 dayguarantee and operate from standard voltages and are of stand-ard size. All are IBM-PC compatible (if 31/2' supported).3.5" Panasonic JU363/4 720K or equivalent £29.95(B)3.5" Mitsubishi MF355C-L. 1.4 Meg. Laptops only' £29.95(B)3.5" Mitsubishi MF355C-D. 1.4 Meg. Non laptop £29.95(B)5.25" EXTRA SPECIAL BRAND NEW Mitsubishi MF501B

360K. Absolutely standard fits most computers £22.95(B) Data cable included in price.

Shugart 800/801 SS refurbished & tested £175.00(E)Shugart 851 double sided refurbished & tested £275.00(E)Mitsubishi M2894-63 double sided switchable

hard or soft sectors- BRAND NEW £250.00(E)Dual 8" drives with 2 mbyte capacity housed in a smart casewith built in power supply' Ideal as exterior drives! £499.00(F)End of line purchase scoop! Brand new NEC D2246 8" 85megabyte of hard disk storagel Full CPU control and industrystandard SMD interface. Ultra hi speed transfer and access timeleaves the good old ST506 interface standing. In mint conditionand comes com.lete with manual. n £299 E

THE AMAZING TELEBOX!Converts your colour monitor into a

QUALITY COLOUR TV!!

TV SOUND& VIDEOTUNER!

The TELEBOX consists of an attractive fully cased mainspowered unit, containing all electronics ready to plug into a hostof video monitors made by manufacturers such asMICROVITEC, ATARI, SANYO, SONY, COMMODORE,PHILIPS, TATUNG, AMSTRAD and many more. The compositevideo output will also plug directly into most video recorders,allowing reception of TV channels not normally receivable onmost television receivers (TELEBOX MB). Push button controlson the front panel allow reception of 8 fully tuneable 'off air' UHFcolour television or video channels. TELEBOX MB covers vir-tually all television frequencies VHF and UHF including theHYPERBAND as used by most cable TV operators. Compositeand RGB video outputs are located on the rear panel for direct n

286 AT - PC286

640k RAM expandable 2 serial & 1 parallelwith standard SIMMS ports

12 Mhz Landmark speed MS-DOS 4.01 20 meg hard disk Co -processor socket 1.2 meg 5-1/4" floppy Enhanced 102 key 1.4 meg 3-1/2" floppy keyboard

Clock & calendar withEGA driver on board batteryback up

BRAND NEW AND BOXED!

orilyE249.00,'The Philips 9CM073 is suggested for the PC286 and rieN

CM8873 for the PC386. Either may use the SVGA MTS-96)0if a suitable card is installed. We can frt this at a cost of £49. )0

òr the PC286 and £39.00 for the PC386.

POWER SUPPLIESPower One SPL200-5200P 200 watt (250 w peak).Semi openframe giving +5v 35a, -5v 1.5a, +12v 4a (8a peak), -12v 1 5a,+24v 4a (6a peak). All outputs fully regulated with over voltageprotection on the +5v output. AC input selectable for 110/240vac. Dims13' x 5' x 2.5'. Fully guaranteed R FE. £85.00 (B)

Power One SPL130. 130 watts. Selectable for 12v (4A) or 24 v(2A). 5v 0 20A. ± 12v 1.5A. Switch mode. New. £59.95(B)Astec AC -8151 40 watts. Switch mode. +5v 0 2.5a. +12v2a. -12v 0.1a. 6-1/4' x 4' x 1-3/4'. New £22.95(B)Greendale 19ABOE 60 watts switch mode.+5v 6a,±12v flla,+15v 1 a. RFE and fully tested.11 x 20 x5.5cms. £24.95',C)Convex AC130. 130 watt hi -grade VDE spec.Switch mode.+5v

15a, -5v 1a,±12v 6a.27 x 12.5 x 6.5cms.New. £49.95 C)Boshert 13090. Switch mode.Ideal for drives & system. +Sy@ 6a,+12v 0 2.5a, -12v 0.5a, -5v 0.5a. £29.95 B)Famell G6/40A. Switch mode. 5v 0 40a.Encased £95.00tC)Famell G24/5S. As above but 24v 5a. £65.0010)

BBC Model B APM Board£100 CASH FOR THE

MOST NOVELDEMONSTRATABLE

APPLICATION!

386 AT - PC386

2 meg RAM expandedby slOts

20 Mhz with 32k cache.Expandable to 64k

40 meg hard disk 1.2 meg 5-1/4" floppy VGA card installed

a

--/MIIMPRIV47

2 serial & 1 parallelports

MS-DOS 4.01 Co -processor socket Enhanced 102 keyboard Kwik Disk Accelerator

Software - FREEBRAND NEW AND BOXED!

onlyE425.00,,MONITORS

14" Forefront Model MTS-9600 SVGAmuttisync with resolution of 1024 x 768.0.28pitch. 'Text' switch for word processing etc.Overscan switch included. Ideal for the PC -386 or PC -286 with SVGA card added. Alsocompatibe with BBC, Amiga, Atari (includingthe monochrome high resolution mode), Ar-

chimedes etc. In good used condition (possible minor screenbums). 90 day guarantee. 15' x 14' x 12'. Only £159(E)

14" Philips Model CM8873 VGA muttisyncwith 640 x 480 resolution. CGA, EGA orVGA, digitaVanalog, switch selectable.

:i,t Sound with volume control. There is also aspecial 'Text' switch for word processing,

''.. spreadsheets and the like. Compatible withIBM PC's, Amiga, Atari (excluding themonochrome high resolution mode), BBC.

Archimedes etc. Good used condition (possible minor screenbums) 90 day guarantee. 15' x 14' x 12'. Only £139(E)Philips 9CM073 similar (not identical) to above for EGA/CGAPC and compats. 640 x 350 resolution. With Text switch withamber or green screen selection. 14' x 12' x 13-1/2' £99(E)KME 10" high definition colour monitors. Nice11tight 0.28' dot pitch for superb clarity andmodem styling. Operates from any 15.625 khzsync RGB video source, with RGB analog andcomposite sync such as Atari, CommodoreAmiga, Acom Archimedes & BBC. Measuresonly 13.5' x 12' x 11'. Also works as quality Tv wan our Hut3Telebox. Good used condition. 90 day guarantee. Only..£125 (E)KME as above for PC EGA standard £145 (E)Brand new C.entronic 14' monitor for IBM PC and compatiblesat a lower than ever pricel Completely CGA equivalent. Hi-resMitsubishi 0.42 dot pitch giving 669 x 507 pixels. Big 28 Mhzbandwidth. A super monitor in attractive style moulded case.Full90 day guarantee. Only ........ ..... £129 (E)NEC CGA 12" IBM-PC compatible. High ,,,quality ex -equipment fully tested with a 90day guarantee. In an attractive two toneribbed grey plastic case measuring 15'L x13'W x 12'H. The front cosmetic bezel hasbeen removed for contractualreasons. Only (E)

20", 22" and 26" AV SPECIALSSuperbly made UK manufacture. PIL all solid state colourmonitors, complete with composite video & sound inputs. Attrac-tive teak style case. Perfect for Schools,Shops,Disco, Clubs.In EXCELLENT little used condition with full 90 day guarantee.

20"....£135 22"....£155 26"....£185 (F)CALL FOR PRICING ON NTSC VERSIONS!

connection to most makes of monitor. For complete compatibility Superb Quality 6 foot 40u- even for monitors without sound - an integral 4 watt audioamplifier arid low level Hi Fi audio output are provided asstandard.Telebox ST for composite video input monitors £32.95Telebox STL as ST but with integral speaker £36.50Telebox MB as ST with Muitiband tuner VHF -UHF -Cable.

& hyperband For overseas PAL versions state5.5 or 6mhz sound specification. £69.95

Telebox RGB for analogue RGB monitors (15khz) £69.95Shipping code on all Teleboxes is (B)

RGB Telebox also suitable for IBM muttisync monitors with RGBanalog and composite sync. Overseas versions VHF & UHF call.

SECAM / NTSC not available.

No Break Uninterruptable PSU'sBrand new and boxed 230 volts uninterruptable power suppliesfrom Densel. Model MUK 0565-AUAF is 0.5 kva and MUD1085-AHBH is 1 kva. Both have sealed lead acid batteries. MUKare intemal, MUD has them in a matching case. Times frominterrupt are 5 and 15 minutes respectively. Complete with fulloperation manuals MUK £249 (F) MUD £525 (G)

1992 Winter Issue: .000. :000.

S. *M. !NOS006 60005 :

BBC Model B type computer on a board. A major purchaseallows us to offer you the PROFESSIONAL version of the BBCcompute' at a parts only price. Used as a front end graphtssystem on large networked systems the architecture of the BBCboard has so many similarities to the regular BBC model B tatwe are sure that with a bit of experimentation and ingenuity manyuseful applications will be found for this board!! It is suppliedcomplete with a connector panel which brings all the I/O to 'D'and BNC type connectors - all you have to do is provide +5 and± 12 v DC. The APM consists of a single PCB with most maj )ric's socketed. The ic's are too numerous to list but include a6502, RAM and an SAA5050 teletext chip. Three 27128EPROMS contain the custom operating system on which wehave no data, On application of DC power the system boots andprovides diagnostic information on the video output. On toadDIP switches and jumpers select the ECONET address andenable the four extra EPROM sockets for user software. App.dims: main board 13' x 10'. VO board 14' x 3'. Supplied testedwith circuit dia ram, data and competition entry form.

9.95 or 2 for £53oSPECIAL INTEREST

Trio 0-18 vdc bench PSU. 30 amps. New £ 470Fujitsu M3041 600 LPM band printer £2993DEC LS/02 CPU board £ 150Rhode & Schwarz SBUF TV test transmitter

25-1000mhz. Complete with SBTF2 Modulator £6500Calcomp 1036 large drum 3 pen plotter £ 650Thuriby LA 160B logic analyser £ 3751.5kw 115v 60hz power source £ 95)Anton Pillar 400 Hz 3 phase frequency converter 75Kw POANewton Derby 400 Hz 70 Kw converter POANikon PL -2 Projection lens meter/scope £750Sekonic SD 150H 18 channel Hybrid recorder £2000HP 7580A Al 8 pen high speed drum plotterKenwood DA -3501 CD tester, laser pickup simulator

£18511£ 355

BRAND NEW PRINTERS

(B)

19" Rack CabinetsMassive Reductions

Virtually New, Ultra Smart!Less Than Half Price!

Top quality 19' rack cabinets made in UKby Optima Enclosures Ltd. Units featuredesigner, smoked acrylic lockable frontdoor, full height lockable half louvered backdoor and removable side panels. Fully ad-justable internal fixing struts, ready

punched for any configuration of equipment mounting plus readymounted integral 12 way 13 amp socket switched mains distribu-tion strip make these racks some of the most versatile we haveever sold. Racks may be stacked side by side and thereforerequire only two side panels or stand singly. Overall dimensionsare 77-1/2'H x 32-1/2'D x 22'W. Order as:Rack 1 Complete with removable side panels £275.00 (G)Rack 2 Less side panels £145.00 (G)

of Display News now available - send large SAE - POCKED with bargains!000.

S.000.0

;

-Es ECTcuii'rf

Microllne 183. NLO 17x17 dot matrix. Full width. £139 (D4Hyundai HDP-920. NU) 24x18 dot matrix full width. £149 (DiOurne LetterPro 20 daisy. Oume OS -3 interface. £39.95 (CY,Centronics 152-2 9 x 7 dot matrix. Full width. £149 (DCentronics 159-4 9 x 7 dot matrix.Serial. 9-1/2' width£ 99 (D.

MAIL ORDER & OFFICESOpen Mon -Fri 9.00-5.30

Dept WW. 32 Biggin Way.Upper Norwood.

London SE19 3XF

LONDON SHOPOpen Mon -Sat 9-5.30Thursday till 9.00pm

215 Whitehorse Lane.South Norwood.London. 5E25

DISTEL 0 The OriginalFree dial -up database!

1000's of items+info on lineV21. V22 & V22 bis

081-679-1888

ALL ENQUIRIES

081167914414Fax- 081-679-1927

All prices for UK Mainland. UK customers add 17.5% VAT to TOTAL order amount. Minimum order £10. PO orders from Goverrvnent,Universities,Schools & Local Authoritieswelcome -minimum account order £30. Carriage charges (A)--42.00. (A1)=C3.75. (B).£5.50. (C)=C9.50. (0)=E11.50. (E)=-£14.00 (F).£18.00 (G)=Call . Scotland surcharge: call.All {pods supplied subject to our standard Conditions of Sale and unless otherwise stated guaranteed for 90 days. All guarantees on a retum to base basis.Rights reserved tochange prices & specifications without prior notice. Orders suteect to stock. Quotations willingly given for higher quantities than those stated. Bulk surplus always wanted for cash.

CIRCLE NO. 128 ON REPLY CARL)

ELECTRONICS WORLD+WIREI ESS WORLD February 1994

INTERFACINGWITH C

by

HOWARD HUTCHINGS

Interfacing with C can be obtained from Lorraine Spindler, RoomL333, Quadrant House, The Quadrant, Sutton, Surrey SM5 2AS.

Please make cheques for £14.95 (which includes postage andpacking) payable to Reed Business Publishing Group.

Alternatively, you can telephone your order, quoting acredit card number. Telephone 081-652 3614.

A disk containing all the example listings used in this book isavailable at £29.96. Please specify size required.

C HERE!If you have followed ourseries on the use of the Cprogramming language, thenyou will recognise its valueto the practising engineer.

But, rather than turning upold issues of the journal tocheck your design for adigital filter, why not have allthe articles collectedtogether in one book,Interfacing with C?

The book is a storehouse ofinformation that will be oflasting value to anyoneinvolved in the design offilters, A -to -D conversion,convolution, Fourier andmany other applications,with not a soldering iron insight.

To complement thepublished series, HowardHutchings has writtenadditional chapters on D -to -A and A -to -D conversion,waveform synthesis andaudio special effects,including echo andreverberation. An appendixprovides a "getting started"introduction to the runningof the many programsscattered throughout thebook.

This is a practical guide toreal-time programming, theprograms provided havingbeen tested and proved. It isa distillation of the teachingof computer -assistedengineering at HumbersidePolytechnic, at which DrHutchings is a seniorlecturer.

Source code listings for theprograms described in thebook are available on disk.

I fsWOK

:411 COMNIUNICATIM----- SERVICES VIA

SATELIJTF_

TOR AND FET

ANUAL

AIL

INSTRUMENTATIONAND TEST GEAR

CIRCUITS MANUAL

Sz.:;

a.

Programmable LogicHandbookGeoff BostockLogic circuit designers areincreasingly turning toprogrammable logic devices as ameans of solving problems. Thisbook, for the establishedelectronics engineer, student andtechnician, is a thoroughintroduction to programmablelogic. Geoff Bostock will take youto a level where you, as adesigner, can take full advantageof the growing product range ofASICs and other self -programmable arrays used incomputer and control systems.Paperback 256 pages.Price £19.95 0 7506 0808 0

Understand Electrical andElectronic MathsOwen BishopPeople who find maths difficultoften have. as a result. difficulty

in grasping electrical andelectronics theory. This book hasbeen written to help suchstudents to understand themathematical principlesunderlying their subject so thatthey can go on with confidenceto tackle problems in practicalcircuits. Paperback 256 pages.Price £14.95 0 7506 0924 9

CIRCUIT MANUALSRay MarstonA series of books dealing withtheir subjects in an easy -to -readand non -mathematical manner,presenting the reader with manypractical applications andcircuits. They are specificallywritten, for the design engineer,technician and the experimenter,as well as the electronics studentand amateur. All the titles arewritten by Ray Marston, afreelance electronics designengineer and international writer.

Op -amp Circuits ManualPaperback 224 pagesPrice £13.95 0 434 912077

Audio IC Circuits ManualPaperback 168 pagesPrice £13.95 0 434 912107

CMOS Circuits ManualPaperback 192 pagesPrice £13.95 0 434 912123

Electronic Alarm CircuitsManualPaperback 144 pagesPrice £13.95 0 7506 00640

Timer/Generator CircuitsManualPaperback 224 pagesPrice £13.95 0 434 912913

Diode, Transistor and FETCircuits ManualPaperback 240 pagesPrice £13.95 0 7506 0228 7

Instrumentation and Test GearCircuits ManualRay MarstonModern instrumentation and testgear circuits of value to theindustrial, commercial, oramateur electronic engineer ordesigner make up this book.Almost 500 outstandingly usefuland carefully selected practicalcircuits are in here. This is onebook you must have if you needaccess to practical workingcircuits ranging from simpleattenuators and bridges tocomplex digital panel meters,waveform generators. and scopetrace doublers. Paperback 400pages.Price £16.95 0 7506 0758 0

Logic Designers HandbookAndrew ParrEasy to read, but none the lessthorough, this book on digitalcircuits is for use by students andengineers and provides anaccessible source of data ondevices in the TTL and CMOSfamilies. It's a 'DesignersHandbook' that will live on thedesigner's bench rather than onthe bookshelf. The basic theoryis explained and then supportedwith specific practical examples.Paperback 488 pages.Price £25.00 0 7506 0535 9

Digital Audio and CompactDisc TechnologyLuc Baert, Luc Theunissen &Guido VergultEssential reading for audioengineers, students and hi-fienthusiasts. A clear and easy -to -follow introduction and includes atechnical description of DAT(digital audio tape). Contentsincludes principles of digitalsignal processing. sampling,quantization, A/D conversionsystems, codes for digitalmagnetic recording. orinciples oferror correction. the compactdisc, CD encoding, opto-electronics and the optical block,servo circuits in CD players,signal processing, digital audiorecording systems, PCM. Video8, R-DAT and S-DAT. Paperback240 pages.Price £16.95 0 7506 0614 2

NEWNES POCKET BOOKSA series of handy, inexpensive,pocket sized books to be kept byyour side and used every day.Their size makes them an ideal'travelling' companion as well.

Newnes ElectronicsEngineer's Pocket BookKeith BrindleyHardback 319 pagesPrice £12.95 0 7506 0937 0

Newnes Electronics AssemblyPocket BookKeith BrindleyHardback 304 pagesPrice £10.95 0 7506 0222 8

Newnes Television aid VideoEngineer's Pocket BookEugene TrundleHardback 384 pagesPrice £12.95 0 7506 0677 0

Newnes Circuit Calct.lationsPocket BookT DaviesHardback 300 pagesPrice £10.95 0 7505 0195 7

Newnes Data CommunicationsPocket BookMichael TooleyHardback 192 pagesPrice £12.95 0 7506 0427 1

Newnes TelecommunicationsPocket BookJE Varrall & EA EdisHardback 400 pagesPrice £12.95 0 7506 0307 0

Newnes Z80 Pocket BookChris RobertsHardback 185 pagesPrice £12.95 0 7506 0308 9

Newnes 68000 Pocket BookMike TooleyHardback 257 pagesPrice £12.95 0 7506 0309 7

Newnes Electrical PocketBook21st editionE A ParrPaperback 526 pages£12.95 0 7506 05138

Newnes Electric CircuitsPocket Book Linear ICRay MarstonHardback 336 pagesPrice £12.95 0 7506 0132 9

Newnes Guide to Satellite TVD J StephensonA practical guide, withoutexcessive theory of mathematics,to the installation and servicing ofsatellite TV receiving equipmentfor those professionallyemployed in the aerial rigging/TVtrades. Hardback 256 pages.Price £17.95 0 7506 0215 5

Newnes Practical RFHandbookIan HickmanPressure on the RF spectrumhas never been greater and itspeople with knowledge and skillsof RF design who are now indemand in the electronicsindustry to design, produce.maintain and use equipmentcapable of working in thiscrowded environment. Thispractical introduction to modernRF circuit design will equip youwith the necessary RFknowledge and skills to enableyou to compete effectively in theindustry. Paperback 320 pages.Price £16.95 0 7506 0871 4

Troubleshooting AnalogCircuitsR A PeaseBob Pease is one of the legendsof analog design. Over the years,he's developed techniques andmethods to expedite the often -difficult tasks of debugging and

troubleshooting analog circuits.Now, Bob has compiled his'battle -tested' methods in thepages of this book. Based on hisimmensely popular series in EDNMagazine, the book contains awealth of new material andadvice for Digital/Analogelectronics engineers on usingsimple equipment totroubleshoot. Paperback 217pages.Price £14.95 0 7506 16326

PC -Based Instrumentation andControlM TooleyDo you need information toenable you to select thenecessary hardware andsoftware to implement a widerange of practical PC -basedinstrumentation and controlsystems? Then this book is foryou. Paperback 320 pages.Price £14.95 0 7506 1631 8

Electronic Circuits HandbookM TooleyProvides you with a uniquecollection of practical workingcircuits together with supportinginformation so that circuits canbe produced in the shortestpossible time and withoutrecourse to theoretical texts.Paperback 345 pages.Price £24.95 0 7506 0750 5

Communication Services viaSatelliteG E LewisDBS is already with us, and willcreate a series of new technicalproblems forengineers/technicians intelevision and communicationservices. This book gives you thesolutions to these problems by:

AUDIO ICCIRCUITS MANUAL

. VARSTON ROBERT A. PEASE

MICROPROCESSORARCHITECTURESAND SYSTEMSRISC CISC &DSF

?11'4:;;;."--"*''

sieve ream

DIGITAL AUDIOAND

COMPACT DISCTECHNOLOGY

explaining how the systemfunctions; describing severalactual systems and givingseveral analyses and designrules. You can't afford to bewithout this invaluabletechnology update if you're asystems design engineer, serviceengineer or technician.Paperback 400 pages.Price £25.00 0 7506 0437 9

Digital Logic DesignBrian HoldsworthAs one of the most successfuland well established electronicstextbooks on digital logic design,this book reflects recentdevelopments in the digital fields.The book also covers newfunctional logic symbols andlogic design using MSI andprogrammable logic arrays.Paperback 448 pages.Price £19. 50 0 7506 0501 4

The Circuit DesignersCompanionT WiliamsThis compendium of practicalwisdom concerning the real -world aspects of electronic circuitdesign is invaluable for linearand digital designers alike.Hardback 320 pages.Price £25 00 0 7506 1142 1

Credit cardordersaccepted byphone

081 652 3614

TroubleshootingAna ogCircuits

Instrumentation & Test Gear Circuits Man 07506 0758 0....16.95Logic Designers Handbook 07506 0535 9....25.00Digital Audio and Compact Disc 07506 0614 2....16.95Newnes Elec Engineers Pkt Bk 0 7506 0937 0...12.95Newnes Elec Assemby Pk Bk 07506 0222 8....10.95

I--Tslewnes TV and Video Eng Pkt Bk 07506 0677 0....12.95171ewnes Circuit Calculations Pkt Bk 07506 0427 1 ....10.95

07506 0308 9....12.9507506 0307 0....12.9507506 0308 9....12.95

Islewnes 68000 Pkt Bk 07506 0309 7....12.9507506 05138 12.9507506 01 32 9 12.95

07506 0215 5 17.9507506 0871 4 16.95

Troubleshooting Analog Circuits 07506 16326 14.9507506 1631 8 14.95

I -Electronic Circuits Handbook 07506 0750 5....24.9507506 0437 9....25.00

D gital Logic Design 07506 05014 19.50

Newnes Data Communications Pkt BkI Newnes Telecommunications Pkt Bk

Newnes Z80 Pkt Bk

I---Newnes Electrical Pk BkI Newnes Electric Circuits Pocket Bk

INewnes Guide to Satelite TV

INewnes Practical RF Handbook

I PC -Based Instrumentation and Control

rReturn to: Lorraine Spindler, Room 1_333, Quadrant House,The Quadrant, Sutton, Surrey SM2 5AS

Please supply the following titles:

Qty Title ISBN Price07506 0808 0....19.95

Understanding Electrical & Elec Maths 07506 0924 9....14.95Op -amp Circuits Manual 0434 912077 13.95Audio IC Circuits Manual 0434 912107 13.95

II--C-MOS Circuit Manual 0434 912123 13.95Electronic Alarm Circuits Manual 07506 0064 0....13.95Power Control Circuits Manual 07506 06908 13.95Timer/Generator Circuits Manual 0434 91291 3....13.95Diode, Transistor & FET Circuits Man 07506 0228 7....13.95

Programmable Logic Handbook

I Communication Services via Satellite

PLEASE ADD £2.50 FOR POSTAGEAdd VAT at local rateNB ZERO RATE FOR UK & EIRE TOTAL

Business purchase: Please send me the books listed with an invoice. I willarrange for my company to pay the accompanying invoice within 30 days. I

will attach my business card/letterhead and have signed the form below.Guarantee: If you are not completely satisfied, books may be returnedwithin 30 days in a resaleable condition for a full refund.

Remittance enclosed £Cheques should be made payable to Reed Book Services Ltd.Please debit my credit card as follows:

Access/Master BarclayNisa Amex Diners

Credit Card No. Exp date

NAME (Please print)

ORGANISATION

STREET

TOWN

COUNTY POST CODE COUNTRY

DATE TELEPHONE NUMBER

SIGNATURE

L

T3000VAT RATES6% Belgium, 25% Denmark, 5.5% France, 7% Gemany, 4% Greece, 4%Italy, 3% Luxembourg, 6% Netherlands, 5% Portugal, 3%, Spain. FORCOMPANIES REGISTERED FOR VAT, PLEASE SUPPLY YOURREGISTRATION NUMBER BELOW (customers outside the EEC should leavethis part blank)VAT NO.If in the UK please allow 28 days for delivery. All prices are correct at time ofgoing to press but may be subject to change.Please delete as appropriate. I do/do not wish to recieve further details abouthooks, journals and information services.Reed Business Publishing - Registered Office - Quadrant Hse The QuadrantSutton Surrey SM2 5AS Registered in England 151537

RF ENGINEERING

USING RF TRANSISTORSBiased view

Norm Dye and HelgeGranberg explain the insand outs of bias circuits,and show how to handletemperaturecompensation. Theauthors also look at usingdevices in pulse mode.From the book RFTransistors: principles andpractical applications.

In bipolar transistors, base current is equal tolc(peak)TFE So the base bias supply must beable to supply this current without too

much effect on the base -emitter voltagebetween the no -signal and the maximum sig-nal conditions. The supply should also be aconstant voltage source, as variations of a fewmillivolts represent a large portion of the nom-inal 0.63-0.67V typical value.

But specific applications demand manyother requirements of the base bias voltagesource.

In some instances a large value capacitorcan be connected across the voltage supply,further reducing its ac impedance. But thismakes impedance dependent on frequency ofmodulation, and is a practical solution onlywhere the modulating frequency is in themedium to high audio frequency range.

One of the simplest biasing circuits for bipo-lar transistors (Fig. 1) uses a clamping diodeto provide a low impedance voltage source.

Forward current of the diode must be greaterthan the peak base current of the transistor. Inthe circuit, current is adjusted with R,) and theresistance of RFC1 and R, is used to reducethe actual base voltage to a slightly lowervalue than the forward voltage of D1.Mechanical connection to the heat sink or thetransistor housing performs a temperaturecompensating function for Tr1 - an adequate

50/10W

Fig. 1. Simple biasing circuit using aclamping diode. It is inefficient since aminimum of liki,k) must go through Pl.Tr, is the transistor to be biased.

81

R2

RFC

Fig. 2. A clamping diode scheme isused as in Fig. 1, but a low currentflows through the diodes and is thenamplified by an emitter follower.

solution although for perfect temperaturetracking, Tr1 and Di should have similar dcparameters.

A disadvantage of the circuit is its ineffi-ciency, especially in biasing high powerdevices: (V-Vb)x/b(max) will always be dis-sipated in the dropping resistors.

But the loss can be overcome by amplifyingthe clamping diode current with an emitter fol-lower (Fig. 2). Two series diodes (D1 and D2)are used so that one can compensate for theVBE(O drop in Tr1. In this case low current sig-nal diodes can be used and their forward cur-rent is equal to I(bia,)/hFE(Tr1)

For best result, Tr1 should have a linear hFEup to the peak bias current required. In higherpower systems Tr1 must be cooled, and ideal-ly, it and one of the series diodes shouldremain at ambient temperature. The otherdiode (D1 or D2) can be used for temperaturecompensation of the rf device.

When the diode (having long leads) is locat-ed near the rf transistor the result is an effec-tive fast -responding system. The leads can beformed to allow the body of the diode to bepressed against the ceramic lid of the rf tran-sistor and fastened in place with thermallyconductive epoxy.

Resistor R1 sets the bias idle current and R2limits its range of adjustment - the value of R2depends on the supply voltage employed.Capacitor C1 and rf choke rfc are there simplyto prevent the rf signal from getting into Tr1.

Another fairly simple bipolar bias source(Fig. 3) has its output voltage equal to the baseemitter junction drop of Tr1 plus the dropacross R3. R1 must be selected to provide suf-ficient base drive current for Tr2, set by itshFE. Normally this current is in the range of afew milliamperes, and Tr1 can be any smallsignal transistor in a package that can be eas-ily attached to the heat sink or rf transistorhousing for temperature compensation. Theonly requirement is that its VBE(O at that cur-rent must be lower than that of the rf transistorat its bias current level. Maximum currentcapability depends on Tr2 and R7.

Power dissipation of Tr2 can be up to a fewwatts. In most cases it should be heat sunk.


RF ENGINEERING

But it must be electrically isolated fromground. The value of R2 can be calculated as:

(VCE-VCE(sat)//b

Capacitors C1_3 are a precaution to suppresshigh frequency oscillations, but may not benecessary depending on the transistors usedand the physical circuit lay -out.

Output source impedances for the circuit,when used in conjunction with a 300W ampli-fier, have been calculated as low as 200-300m11

Biasing mosfetsGate threshold voltages of mosfets are 5-6V,so some gate bias voltage is needed in mostapplications.

Class C operation is possible (zero gatebias), though there is a cost in low power gain:the input voltage swing must be big enough toovercome the gate voltage, from zero to overthe threshold level. But drain efficiency is usu-ally higher than in other classes. Operation canalso approach class D - especially if over-driven.

Zero bias is often used in amplifiers intend-ed for signals not needing linear amplification- such as fm signals and some forms of cwsignals. Efficiencies in excess of 80% are notuncommon.

In class B, the gate bias voltage is set justbelow the threshold, resulting in zero drainidle -current flow. Power gain is higher than in

class C, but drain efficiency is 10-15% lower.Class B is also suitable only for non-linearamplification.

Between classes, the decisions to be madeare whether the system has power gain tospare and the importance of efficiency.

At higher frequencies, such as uhf, a goodcompromise may be class B or even class AB.In class AB the gate bias voltage is somewhathigher than the device threshold, with drainidle -current flow resulting.

The idle current required to place the devicein the linear mode is usually given in a datasheet. In this respect, mosfets are much moresensitive to idle current than bipolar transis-tors, and also require somewhat higher currentlevels compared to bipolars of similar electri-cal size.

Temperature compensationTemperature compensation of mosfets canmost readily be accomplished with networksof thermistors and resistors - the ratio of thetwo must be adjusted for thermistor charac-teristics and the gf, of the fet. Changes in thegate threshold voltage are inversely propor-tional to temperature and amount to approxi-mately ImV/°C. They have a larger effect onthe /DQ of a fet with high gfs than one withlow gfs. Unfortunately the situation is com-plicated by the fact that gf, is also reduced atelevated temperatures, making the drain idle -current dependent on two variables.

In spite of this dependence, this compensa-

Operating transistors in a pulse modeI\F energy, in the form of pulses, is exploited in manyapplications including medical electronics, laser

excitation and various types of radar. In each, thespecifications of carrier frequency, pulse repetition rate andduty cycle vary.

The carrier frequency is usually much higher than the pulserepetition rate, resulting in the generation of bursts of rf at thecarrier frequency whose lengths depend on the pulse width.Pulse repetition rates are typically in the audio range andduty cycles range between 0.05 and 10%.

For low duty -cycle applications - such as radar - specialdevices have been developed to operate at higher peakpowers, but with relatively low average powers, reducingdissipation. These transistors (uhf to microwave) are almostexclusively bipolar.

In general, bjts have higher peak power capabilities thanmosfets. Peak power performance can be further improvedby reducing the emitter ballast resistor values to lower thannormally required for cw.

The epitaxial layer that controls the transistor's saturatedpower, is also made thinner than normal since the problemof ruggedness is partly eliminated by the low average power.

Increasing the pulse width increases the dissipation, and atpulse widths of 1 ms and wider the device can be consideredto operate like a cw signal: the temperature time constant ofa medium size rf power die is around 1 ms, beyond whichmore heat will be transferred into the bulk silicon andthrough it to the transistor housing.

If the pulses are short but the repetition rate approaches1 kHz (1ms period), the effect is the same. Transistors made

C3

Fig. 3. Bias circuit featuring the lowestsource impedance of the less complexbias circuits. It is recommended for highpower device biasing and for otherdemanding applications.

tion method can operate satisfactorily and isrepeatable for production. The thermistor isthermally connected into a convenient locationin the heat source in a similar way to the com-pensating diodes with bipolar units. Figure 4shows a simple mosfet biasing circuit.

Most mosfet device data sheets give Vgs(tri)versus /D data. But the values are only typical.In some cases gfs can vary as much as 100%from unit to unit. In production, the devicesshould have gfs values within 20% of eachother, or every amplifier will have to be indi-

exclusively for pulse operation can produce peak powerlevels of five to six times the cw rating for a die of a similarsize.

Standard transistors designed for cw have a multiplyingfactor more of the order of three to four.

Mosfets for pulsed powerMosfets can be used for pulsed power operation, but theyhave disadvantages as well as advantages compared to bjts.Disadvantages include pulse drooping, where the trailingend of the pulse has a lower amplitude than the leading end.It is caused by the decreasing gfs of a mosfet withtemperature. Corrective circuitry can compensate, but addsto circuit complexity.

Advantages include smaller phase delays and faster riseand fall times. Choice depends on the application and onwhat the designer decides which is the most suitable.

Additional considerations in design of a pulsed amplifierare energy storage near the device, and minimisinginductance in the emitter leads. Both affect rise time of thepulse and prevent droop resulting from voltage decay duringits duration.

Some trade-off will be required because as the emitterinductance to ground is reduced, wideband matching ismade more difficult. Also a minimum amount of inductanceis needed in the collector circuit to achieve adequatedecoupling. But pulses with rise times of the order of tens ofnanoseconds can be obtained with devices that deliver up toseveral hundred watts of power over bandwidths of at least20 to 30%.


RF ENGINEERING

R6 C2

Fig 4. Mosfet bias circuit using a thermistor-res'stor network for temperaturecompensation.

vidually checked for temperature tracking.Some manufacturers such as Motorola supplyrf power fets with specified ranges of gf,matching.

A closed loop system for mosfet biasing(Fig. 5) could provide an automatic and pre-cise temperature compensation for any mosfetregardless of electrical size and gfs.

No temperature sensing elements need beconnected to the heat sink or to the devicehousing. In fact, fets with different gatethreshold voltages can be changed in theamplifier without affecting the idle current, sothe gate threshold voltage can vary withinwide limits over a short or long time for avariety of reasons.

Other factors affecting Vgs(th), in addition totemperature, might be moisture levels andatmospheric pressure.

Principle of automatic biasIn the automatic bias circuit, the idle current ofthe mosfet amplifier is initially set to class A,AB, or anywhere in between these bias limitsby R7 - which also provides a stable voltagereference to the negative input of the opera-tional amplifier U1.

Current flows through R1 with a consequentvoltage generated across it. The voltage is fed

R1

R2

VDS

to the positive input of U1, resulting in the out-put of Ui following it in polarity but not inamplitude. Due to the voltage gain in U1,which operates in a dc open loop mode, itsoutput voltage excursions are much higherthan those generated across R1. So if the cur-rent through R1 tends to increase for any rea-son, part of the output voltage of U1 fed to theamplifier gate bias input will adjust to a lowerlevel, holding the current through RI at itsoriginal value. A similar self adjustment willalso take place in the opposite direction.

Values for the resistive voltage divider R4..5have been selected for a suitable range, suffi-cient to control the amplifier fet gate with thefull voltage swing at the output of U1.

When the amplifier is rf driven, the currentthrough R1 increases and the bias voltage tothe amplifier tends to decrease along with thevoltage to the positive input of U1.

But at the same time, Tr1 will start conduct-ing, lowering the effective value of RI sinceTr1 is in parallel with it. The turn -on gate volt-age for Tr1 is obtained from the voltage dropacross R2.

Typical values for R1 are 5-100 and for R2,0.1-0.2a The values must be selected on thecharacteristics of Tri, the exact application andthe currents in question.

The higher the current drawn by the ampli-fier, the harder will Tr1 be turned on. Forexample if R1 is 552 and Tri is fully turned onwith its rDs(on) of 0.212, the effective value ofR1 will vary between 552 and less than 0.212,depending on the current drawn. So the cur-rent variable resistor (Tri-R1) makes it possi-ble to keep the output of U1 and the resultingamplifier bias voltage relatively stable undervarying current conditions.

The circuit is ideal for class A amplifiers,where the drain current remains constantregardless of the rf drive. Tr1, R2 and R3 canbe omitted for class A amplifiers and the valueof R1 can be made as low as 0.05-0.112.

C3

BIASAdjust

Fig. 5. Automatic biastracking system formosfet power amplifiers.It provides automatictemperaturecompensation withoutsensors as well asversatility for substitutinga variety of electricalsizes of fets operating atany supply voltage.

Radio FrequencyTransistorsPrinciples and Practical Application

Norm Dye and Helge Granberg

EON s AtS M

Norm Dye is Motorola's productplanning manager in theSemiconductor Products Sector,and Helge Granberg is Memberof Technical Staff, RadioFrequency Power Group(Semiconductor Products) atMotorola. Their rf transistorsbook includes practicalexamples from the frequencyspectrum from 2MHz tomicrowaves, with specialemphasis on the UHFfrequencies .

RF Transistors: Principles andpractical applications isavailable by postal applicationto room L333 EW+WW,Quadrant House, TheQuadrant, Sutton, Surrey, SM25AS.Cheques made payable to ReedBooks Services. Credit cardorders accepted by phone (081652 3614).288pp HARDBACK 07506 90593 Cost f19.95 + Postage £2.50

1 72 ELECTRONICS WORLD + WIRELESS WORLD February 1994

CLASSIFIEDTEL 081 652 8339 FAX 081 652 8931

ARTICLES FOR SALE

Cooke InternationalSUPPLIER OF QUALITY USED

TEST INSTRUMENTSANALYSERS, BRIDGES, CALIBRATORS,

VOLTMETERS, GENERATORS, OSCILLOSCOPES,POWER METERS, ETC. ALWAYS AVAILABLE

SPECIALIST REPAIR WORK & CALIBRATIONUNDERTAKEN

ORIGINAL SERVICE MANUALS FOR SALECOPY SERVICE ALSO AVAILABLE

EXPORT, TRADE AND U.K. ENQUIRIES WELCOME,SEND LARGE "A3" S.A.E. + 50P POSTAGE FOR LISTS

OF EQUIPMENT AND MANUALS.ALL PRICES EXCLUDE VAT AND CARRIAGE

DISCOUNT FOR BULK ORDERS SHIPPING ARRANGED

OPEN MONDAY -FRIDAY 9AM-5PM

Cooke InternationalELECTRONIC TEST & MEASURING INSTRUMENTSUnit Four, Fordingbridge Site, Main Road, Barrham,

Bognor Regis, West Sussex, P022 OEB

Tel: (+44) 0243 545111/2 Fax: (+44) 0243 542457

HIGH END TEST & COMMUNICATIONSEQUIPMENT PURCHASED


TURN YOUR SURPLUSTRANSISTORS, ICS ETC, INTO CASH

Immediate settlement.We also welcome the opportunity to quote for

complete factory clearance.Contact:

COLES-HARDING 8 CO. 103 South BrinkWisbech, Cambs PE14 ORJ.

ESTABLISHED OVER 15 YEARSBuyers of Surplus Inventory

Tel: 0945 584188 Fax: 0945 475216

CIRCUIT BOARDS

TELEKINETIk SR5232 CONTROLLED I/O FOR IBM COMPATIBLESINEXPENSIVE DATA ACOUIsr-IOP AND CONTROL

Single board kit composes:Populated PCB with prototyping are: applications with

source code in Turbo Pasca include.A'S Conversion. D/A Conversion, Ste spoor Motor Control,

VFET Applications Wavaorm Synthesis.Pnce of kit including documentation E100 (all inclusive) Allow

21 days for delverySole Proprietor: B. Teleld

(make cheques payable to d TeleIM8 Victoria Court, Brampton Road,

Maybank Newcastle -Under -Lyme, Staffs ST5 OSL.Telephone (0782)662099

FREE CLASSIFIED

SELLING NEW PORTABLE CB tras-ccivcr curosonic FM, P.L.L. 40 channels4/1 watts, UK -model, ultra -slim (7.24H x2.76W x 1.730 inch) squelch/tone con-trol, jack for external micro/earphone/power supply, BNC antenna connector,rubber antenna: Tel: 01043/512/260371(after 8pm, please ask for Christoph).

FOR SALE HP 8556A HP8553B plug-inunits for 141T O.I.R.O. £550. Tel: 0442842457 wanted parabolic antenna andfeed > 1.2m good for 12G.

WHY NOT FAX YOURADVERT ON 081 652 8931

COMPELEC1994 BARGAIN LIST

THE ELECTRONIC ESSENTIAL,MILLIONS OF COMPONENTS AT

UNBEATABLE PRICES.FOR A FREE COPY PHONE,

FAX, OR WRITE (TRADE ONLY)

COMPELEC14 CONSTABLE ROAD, ST. IVES,

HUNTINGDON, CAMBS. PRI7 6EQ

0480 300819WE ALSO PURCHASE ALL TYPES OF

ELECTRONIC COMPONENTS. LISTS T/ABOVE ADDRESS.

ARTICLES WANTED

WANTEDTest equipment, receivers, valves,transmitters, components, cable

and electronic scrap and quantity.Prompt service and cash.

M & B RADIO86 Bishopgate Street,

Leeds LS1 4BBTel: 0532 435649Fax: 0532 426881 3956

WANTEDReceivers, Transmitters, Test

Equipment, Components, Cabeand Electronic, Scrap. Boxes,

PCB's, Plugs and Sockets,Computers, Edge Connectors.

TOP PRICES PAID FOR ALL TYPES OFELECTRONICS EQUIPMENT

A.R. Sinclair, Electronics, Stockholders,2 Normans Lane, Rabley Heath, Welwyn,Herta ALS 9T0. Telephone: 0438 812193.

Mobile. 0860 214302. Fax: 0438 812 387Telephone: 0763 246939 780

STEWART OF READING110 WYKEHAM ROAD,READING, RG61PL.TEL. 0734 268041FAX: 0734 351696

TOP PRICES PAID FOR ALLTYPES OF SURPLUS TESTEQUIPMENT, COMPUTER

EQUIPMENT,COMPONENTS, etc.

ANY QUANITYT103

WANTEDHigh -end Test Equipment, only brand

names as Hewlett-Packard,Tektronix, Rhode & Schwarz, Marconi

etc. Top prices paid.Please send or fax your offer to:

HTB ELEKTRONIKAlter Apeler Weg 5, 2858 Schiffdorf,

West GermanyTEL: 01049 4706 7044FAX: 01049 4705 7049

ARTICLES FOR SALE

NEW VMEDSP MPV901 Boxed. Offers:0703 733312.

WANTED VALVES especially KT66,KY88. PX4, PX25 (also transistors IC'scapacitors, valve radios/hi-fi). If possiblesend written list for offer by return toBillington Export, 1E Gilmans IndEstate, Billinghurst, Sussex, RH14 9EZ.Tel 0403 784961. Fax 0403 783519.

WE WANT TO BUY !!

IN VIEW OF THE EXREMELYRAPID CHANGE TAKING PLACE

IN THE ELECTRONICSINDUSTRY, LARGE QUANTITIES

OF COMPONENTS BECOMEREDUNDANT. WE ARE CASH

PURCHASERS OF SUCHMATERIALS AND WOULD

APPRECIATE A TELEPHONECALL OR A LIST IF AVAILABLE.

WE PAY TOP PRICES ANDCOLLECT.

R.HENSON LTD.21 Lodge Lane, N.Finchley,

London N12 8JG.5 Mins, from Tally Ho Corner.

TELEPHONE081-445-2713/0749FAX 081-445-5702.

1

ADVERTISERSPLEASE NOTEFor all your future

enquiries on

advertising rates,

please contact

Pat Bunce on:

Tel: 081-652 8339Fax: 081-652 8931

BOOKS & MANUALS

THE VINTAGE WIRELESS BOOK LISTINGPublished regularly containing 100s of out.of.print, old and collectable wireless and TV books, amateur

radio books and magazines. Send six first class stamps for 100 page listing or £3.75 for next four issues.

FPIEW 1100KS

LANES MILITARY COMMUNICATIONS 1989. 10th Edition. A vast volume (862 pages). Large format wraps.

Contains descriptions, photographs and basic technical details of the world's military communications

equipment. Brand new in carton. Published at £80 Special Offer145 including postage U.K. Foreign

postage extra.

JANES COMMAND INFORMATION SYSTEMS 1989-1990.(C31 Systems). Large format 208pp. Contains

descriptions, photographs and basic technical details of the world's military command communications and

intelligence gathering systems, including radar equipment. Brand new in carton. Published at £80. SpecialOffer £40 inclusive of postage U.K.

WINNING THE RADAR WAR A new book on World War 2 radar. The suspense filled story of the experiments

and electronic eavesdropping. Author was one of the key technicans. 224pp illus £9.95 + £2.00 p+ p

CHEVET VINTAGE SUPPLIESDept P.W. 157 Dickson Road, Blackpool FY1 2EU

Tel: 0-253 751858 or 0253 302979VISA

REPRINTSa ready made sales aidIf you are interested in a particular article oradvertisement, you should take advantage ofour reprint service. We offer an excellent,reasonably priced service. For further detailsand a quotation (minimum no. 250), contact:

Jan Crowther, Room 1006,Quadrant House, The Quadrant,Sutton, Surrey SM2 5AS, UKTelephone: 081-652 8229Fax: 081-652 4728

CLASSIFIED ADVERTISEMENT ORDER FORM

1 2 3 4 5 6

7 8 9 10 11 12

13 14 15 16 17 18

19 20 21 22 23 24

25 26 27 28 29 30

Place a lineage advertisement in next month's issue and it will cost,for a single insertion, only £2.10 per word.

Special rates:6 insertions £2.10 per word/issue(Advertisement can appear everymonth or every other month only).

WHY NOT PLACE A BOXED AD-VERTISEMENT TO GIVE MAXIMUMIMPACT?

Extras:Spot Colour 20%

Box number service £22.00

EXAMPLESIZE

3cmx 1 columnFor 1 insertion cost

is: £45.00

Lineage advertisements under £50 have to be pre -paid by credit cardor cheque.

ALL RATES QUOTED ARE EXCLUSIVE OF VAT:All major credit cards accepted

Please debit my card a total of E

Expiry DatePlease ensure that address given is where your credit card statement goes to.

NAME

ADDRESS

TEL NO SIGNATURE

All advertisements must be received five weeks prior to publication date.All cancellations must be received by eight weeks prior to publication date. Afterthat no advertisement can be cancelled.

Please send to Electronics World & Wireless World, Classified, 11th Floor,Quadrant House, The Quadrant, Sutton, Surrey SM2 5AS. Tel: Pat Bunce on081-652 8339.

ELECTRONIC UPDATEContact Pat Bunce on

081-652 8339

OD C:1 Non.L4 nartm

Model, 82200 and 82400

Gang and Set Programmers for24, 28 8 32 pm EPROM.. EEPROMs,

FLAW Emulators Ind OTPs up to 8M bit.

The system 200(1 is an idealprogrammer for the produc-tion environment. Fiat prog-ramming results in highthroughput and rigonws veri-fication leads to improvedquality control. Single keyfunctions and checks againstmisoperation facilitates itsuse by unskilled staff.

MQP ELECTRONICS LTD.Tel: 0666 825146Fax: 0666 825141


OLSON ELECTRONICS L'MITED isa leading manufacturer in tie field ofmains distribution panels of everyshape and size to suit a va iety ofneeds. For use in Broadcasting,Computing, Data Communications,Defence, Education, Finance,Health etc. All panels aremanufactured to BS5733. BRITISHAMERICAN, FRENCH, GERMANCEE22/IEC and many othersockets. Most countries catered for.

All panels are available ex -stock andcan be bought direct from OLSON.

Olson Electronics LimitedTel: 081 885 2884Fax: 081 885 2496


41*.,i;7117.711b1

7:7-44

1.001111. P.O.MGOESTER0113.,0.00 1161 .173 62S1

Oa! rt^...a..- . 99

IR Group, Europe's leadingsupplier of used

instrumentation, haspublished the latest updateof models available. With arange from power suppliesto network analysers, mostitems are available on short

delivery and come with a12 month parts and labourwarranty. For a detailed

quotation call0753 670000.


ELECTRONICSWORLD+ WIRELESS WORLD

Cr:7MM(hap osrInnis Judo

design,

watingstithswikiledopariior=Mininrerialwirelea sysiern

e Tess discredit6n Iheon

ELECTRONIC UPDATE isElectronic World andWireless World's section foradvertisers to market theirproduct information. Fromcatalogues to newsletters,Data Update is designed topresent your productinformation in a clear andattractive manner while our"CIRCLE NUMBERS"help readers to obtain theinformation they need fast.


A regular advertising feature enablingreaders to obtain more informationon companies' products or services.

FREE VXI BROCHURE

The National InstrumentsVXI brochure describes thecompany's embedded PC andGPIB controllers, MXIbusinterface kits for multipleplatforms, and NI-VXI,LabWindows, and Lab VIEWsoftware for developing andcontrolling VXI instrumen-tation systems.

NATIONAL INSTRUMENTS

Tel: 0800 289877CIRCLE NO. 146 ON REPLY CARD

DATAUPDATE isElectronics Weekly'ssection for advertisers tomarket their productinformation. Fromcatalogues to newslettersData Update is designed topresent your productinformation in a clear andattractive manner, whilstour colour coded enquirynumbers help readers toobtain the information theyneed fast.CIRCLE NO. 147 ON REPLY CARD

2nd EDITION TOKO RFCATALOGUE

Cirkit have just published the 2ndEdition of the Toko RF Catalogue,featuring details of Tokos' extensiverange of RF coils, inductors, filtersand comms ICs.

The 128 page catalogue includesmany new products such as;Surface mount high currentinductors, surface mount multilayerinductors, helical filters at 2.5GHzand a new section of push buttonand tact switches.Cirkit Distribution Ltd, Park Lane,Broxbourne, Herts, EN107NQTel: (0992) 441306Fax: (0992) 441306

CIRCLE NG. 148 ON REPLY CARD

High Speed EPROM & FLASHProgramming from your PCO Programs EPROMs to 4 Mbits/

32 -pinsO Superfast 8, 16 & 32 -bit

programmingCI Approved algorithms Menu driven software includedEl Sophisticated editor functionsCI Easy file managementO FREE demo disk available

Stag Programmers LimitedMartinfield Welwyn Garden City,Herlordshire, AL7 1 JT UKTel: (0707) 332148Fax: (0707) 371503


SPECTRUM ANALYSERS

U1111111116:* HP85698 10MHz-226Hz MICROWAVE SPECTRUM ANALYSERS *

High -Performance features include external mixing to115GHz, Internal Preselection

1 .7-22G Hz. Wide resolution range 100Hz-3MHz. On -screen readout,direct pbtter output. Digital storage. RF Input from -123 to + 30dbm. HP1B.

'UNREPEATABLE' Price £4950 each

* MARCONI TF2370 1250MHz SPECTRUM ANALYSER SYSTEM *

TF2370 including TK2373 frequency extender to 1.25GHz, 100dbdynamic range. 0.1db & 5Hz resolution. Frequency resolution to 1Hz.

Many features inc phase lock, digital storage, tracking generator.Modern brown livery. Factory New condition £2250

MARCONI INSTRUMENTS2017 signal generator 10kHz-1024MHz

microprocessor -controlled cavity -tuned low -nose

AM/FM, +164dbm output. Few only FREE CALIBRATIONINCLUDING FREE CALIBRATION -£2000 each.

2019 sysnthesized AM/FM signal generator 80kHz-1040MHz

6059A signal source 12-18GHz

6140 GPIB adapter

6460/6420 power meter 10MHz-12.4GHz 0.3uW-10mW

6460/6423 power meter 10MHz-12.4GHz 0.3mW-3W

6700B sweep oscillator 8-12.4GHz & 12.4-18GHz

6960/6910 digital RF power meter 10MHz-20GHz GPIB

6912 power sensor 30kHz-4.2GHz for above series

8936 audio power meter

OA2805A PCM regenerator test set

TF2370110MHz spectrum analyser

TF2910/4 non-linear distortion (video) analyser

TF2914A TV insertion signal analyser

TF2910 TV interval timer

EASTCOTE LANE S MIDDLESE \ HA, SDBTEL; 081-422 3593. FAX: 081-423 400.9

NOWIN

40thYEAR

01ST .11-111416/. ASS

TEST EQUIPMENTANRITSU ME518A PCM portable error rate test setBRUEL & KJAER 2511 vibration meter seV1621 filterBRUEL & KJAER 2610 measuring amplifier

BRUEL & KJAER 2307 level recorderBRU EL & KJAER 2317 portable level recorderBRUEL & KJAER 1618 band bass filterBRUEL &KJAER 3204 tapping machineBRUEL & KJAER 3513 portable vibration analyserBRUEL & KJAER 2515 vibration analyserAVORM215L-2 insulation & breakdown testerDATRON 1065 digital muttimeter

DRANETZ 626 mains disturbance analyser/2 x PA -6001DRANETZ 606-3 line disturbance analyserFLANN MICROWAVE 27072 frequency meter 73-113GHzFLUKE 5100A calibratorKEITHLEY 192 programmable dmm

£3000£2250

000E- 000

£-850750

E1000

£:500E.9)00

350E750

£1250£275

£275E6000

£400

MUCH MORE. FULLY REFURBISHED FULLY GUARANTEED TESTEQUIPMENT AVAILABLE FROM STOCK PLEASE REQUEST OUR

CURRENT LISTINGS WE CAN FAX LISTS & SHIP GOODS WORLD-WIDE HIGH END EQUIPMENT ALWAYS WANTED FOR STOCK

CALL US NOW

£1950

£750

£100

£350

£400

£1000

£1000

£150

£350

£750

£1750

£1000

£1250

£500

MAURY MICROWAVE 8650E INC-calibration AtNAGRA IV -SJ tape recorderPHILIPS P5193 synthesized function generatorPHILIPS PM2534 digital multimeterPHILIPS PM8272 XY & Xt dual -pen analogue plotterRACAL 2101 frequency counter 21GHzRACAL 9008 automatic modulation meter £325.9009RACAL 9081 synthesized AM/FM sig' gen' 5-520mHzRACAL 9300 RMS voltmeter -80dB to +50dBRACAL 9341 LCR databridgecomponent testerRACAL -DANA 9302 RF milli -voltmeter 1.5GHzRACAL -DANA 9303 level meter, digital

ROBERTS &ARMSTRONG I/o -cable end -cut measure' unitTEKTRONIX TM503/SG503/TG501/PG506 scope calibratorTEKTRONIX J16 digital photometerTEKTRONIX AM503 current amplif ier/TM501 /A630220A probe

TEKTRONIX AM503/A6303/TM501 100A current probeTEKTRONIX 2235100MHz oscilloscopeTELON IC 1205A 1-1500MHz sweep generatorWAVETEK 20000-1400MHz sweep generator

0530E2030E1530£400£550

£16,0

£600

£325£350

£43)£100)

£500£225)

£25)£125)

£1500E951

MOH£751

HEWLETT PACKARD

331A distortion meter339A distortion meter3406A sampling voltmeter355C attenuator DC-1GHz 0-11db & 355D 0-120db

3325A synthesizer/function generato3335A synthesizer/level generator with option 01

3552A transmission test set

£200£1500£250£125

£1500

£1500£1250

3562A dynamic signal analyser £700037I1A/3712A microwave link analyser (MLA) with 37938 & 3730B/3736B RFdown -converter (1.7-4.2GHz) £40003781A pattem generator £12503782A error detector £1250400FL mV -Meter 100uV-300V fs. 20Hz-4MHz £325415E swr meter £3504274A muki-frequency (100Hz-100kHz) LCR component meter £40004276A/001 LCZ meter £1750432A/478A microwave power meter 10MHz-1 OGHz £400432A/R486A uwave power meter 26.5-40GHz (waveguide) £6005342A 18GHz frequency counter £12505343A 26GHz frequency counter E22506253A dual power supply 0-20V 0-3A twice £2256825A bipolar power supply/amp -20 to +20vdc 0-1A £35070300A tracking generator plug-in unit £200070907A external mixer for 70000-ser spectrum analyser £175070358 X -Y single pen analogue chart recorder £3507790 dual -directional coupler 1.7-12.4GHz (also others) £3508112A pulse generator £2508011A pulse generator 0.1Hz-20MHz £500816A slotted lion 1.8-18GHz with carriage 809C & 447B E5008405A vector voltmeter, voltage & phase to 1000MHz £9508406A comb generator £10008447A RF amplifier 0.1-1300MHz, 22db gain, 0.1W o/p £5008505A network analyser system including 8503A S -parameter test set and8501A storage normaliser £50008601A 110MHz sweep generator £5008620C sweep generator, many plug-in units available £call8671A synthesized signal generator 2-6.2GHz £25008673 2-18GHz synthesized signal generator £7500

PLEASE NOTE: ALLOUR EQUIPMENT IS NOW OPERATION -VERIFICATION

TESTED BEFORE DESPATCH BY INDEPENDENT LABORATORY

We would be pleased to handle all grades of calibration or NAMAS certification

by same laboratory at cost price. All items covered by our 90 -day parts and labour

guarantee and 7 -day 'Right to Refuse' (money back) warranty.

ALL PRICES SUBJECT TO ADDMONAL VAT AND CARRIAGE


INDEX TO ADVERTISERS

Amplicon Liveline

Bull Electrical

Citadel Products LtdChemar Value

Dataman Programmers Ltd

PAGE

143

95

IFC150

OBC

M & B Radio (Leeds)MQP Electronics

Number One Systems

Pico Technology LtdPowerware

PAGE

135

145

155

136

96Display Electronics Ltd 166

Electrovalue Ltd 135 Ralfe Electronics 176

Ericsson 155 Research Communications 96

Field Electric 145 Seetrax Ltd 111

Halcyon Electronics Ltd 145 Smart Communications 136, 145Stewart of Reading 150

Johns Radio 99 Surrey Electronics Ltd 150JPG Electronics 150

Kestral Electronics 136 Telnet 136

Keytronics 116 Texas Instruments 90

Kinloch 135 Tsien Ltd 90

Labcentre 106 Ultimate Technology IBC

176 ELECTRONICS WORLD+WIRELESS WORLD February 1994

FROM CONCEPT TO ARTWORK IN 1 DAY

sign capacity 350 pins

e Incl. S & H. excl. VAT: 7-

-6::

g- ideas are quickly captured using the ULTIcap schematic design Tool. ULTIcap usesME checks to prevent logic errors. Schematic editing i; painless: simply click your start and end

ind JLTlcap automatically Awes them for you. ULTIcap's auto snap to pin and auto junctionfeatures ensure your netlist is complete. thereby relieving you of tedious netlist checking.

L -H- ell. :he integrated user interface. makes sure alln sign information is transferred correctly from

tc ULTIboard. Good manual placement tools arevital to the progress of your design. therefore ULTlboardgives you a powerful suite of REAL-TIME functions suchas. FORCE VECTORS. RATS NEST RECONNECT andDENSITY HISTOGRAMS. Pin and gate swapping allowsyou to further optimise your layout.

No.v you can quickly route your critical tracks.UiLTIocard's REAL-TIME DESIGN RULE CHECKw!l, not allow you to make illegal connections orvolate your design rules. ULTIboard's powerfulTRACE SHOVE, and REROUTE -WHILE -MOVEalgorithms guarantee that any manual trackediting is flawless. Blind and buried vias andsurface mount designs are fully supported.

If you need partial ground planes, thenwith the Dos extended board systems youcan automatically create copper polygonssimply by drawing the outline. The polygonis then filled with copper of the desired net.all correct pins are connected to thepolygon with thermal relief connections

and user defined gaps are respectedaround all other pads and tracks.

ULTIboard's autorouter allows you tocontrol which parts of your board areautorouted, either selected nets. or acomponent, or a window of the board, orthe whole board. ULTIboard's intelligentrouter uses copper sharing techniques tominimise route lengths. Automatic viaminimisation reduces the number of vias todecrease production costs. The autorouterwill handle up to 32 layers, as well assingle sided routing.

ULTIboard's backannotation automaticallyLpdates your ULTIcap schematic with any pinand gate swaps or component renumbering.Finally, your design is post processed to generatepen / photo plots, dot matrix/laser or postscriptprints and custom drill files.


ULTIboard PCB Design/ULTIcap

Schematic Design Systems are available

in low-cost DOS versions. fully compatible

with and upgradable to the 16 and 32 bit

DOS -extended and UNIX versions.

featuring unlimited design capacity.

4Etvivem. 14.4144, 4044.4,z.v.t

ULTImate Technology UK Ltd. 2 Bacchus House, Calleva Park, Aldermaston Berkshire RG7 4QW Fax: 0734 - 815323 Phone: 0734 - 812030

The legendary S4 -

the smallest, most powerfulpersonal programmer you

pore ,=&11000400.r:tryn buy - and only Y./495!,1

'TOM engine

management to

Antarctic survey teams,

you can find S4s the

world over, up and

running where the

competition is left far

behind. S4 gets the job

done in every situation

you might expect - and

quite a few you wouldn't!

1.41We.'CAPABILITIES

A 32 pin ZIF socket programs a

huge library of EPROMs,

EEPROMs and FLASH devices.

Dataman S4 programs devices up

to BMbits and the unique,

loadable Library means that new

parts can be added quickly without

extra cost! Serial EPROMs, 40 pin

EPROMs and micro -controllers

are all supported with optional

modules.

eoi0000w°°°°°n

VW,

-=,,...,..3,-...-----::

,s..,.-.--,

.c -

--

:r

k -;;;1,

SNUFF/\'0t01;'

.,,.V. "o

'' 9o,.

, ;`,

\' '0.'' ;

1'

°921

\'\''' .".'.. -. :,..-`.'`. k.....\we,.

ea.

\ :1;;';.' ''''' _. '....... 4..... ''.." ,

W.

... ,,;10,

.o.

..:... *e*".'

Size:

186 x 111 x 46mmWeight: 515g

11,

,`; 4.

v.*

Dataman Programmers Ltd

Credit Card Hotline:(0300) 320719

for same -day dispatch

'c4 EMUI 41'!riN

With Dataman S4's powerful

emulation system you see your

code running before committing

yourself to an EPROM. Simply

download your code to S4, press

'EMULATE', and your target

system runs, in real time, as if an

EPROM was plugged in.

--S4 REMOTE CONTROL

Dataman S4 has its own internal

processor and memory, but can

also be operated remotely from

your PC at speeds up to 115,200

Baud. S4 is supplied with a freti--

disk containing custom terminal

software and a pop-up TSR com-

munications

Datamah 34 is shipped ready to

use, complete with a mains

charger, emulation lead, write

lead, personal organiser/

instruction manual, MS-DOS

communications software, spare

Library ROM - and a 3 year

guarantee.

Station Road, Maiden Newton, Dorset. DT2 OAE, UK. Telephone: 0300 320719 Fax: 0300 321012 Telex: 418442 BBS: 0300 321095 24hr Modem: V32bis/16.8K HST.

22 Lake Beauty Drive, Suite 101, Orlando. FL 32806. USA. Telephone: (407) 649-3335 Fax: (407) 649-3310 BBS: (407) 649-3159 24hr Modem: V32bis/16.8K HST.

CIRCLE N0_131 ON REPLY CARD

Documents

ELECTRONICS WORLD · 2019-07-17 · 111111111111111[11 11111 Ill 11111111111111111111 t rrupt m/NIor e, pan a Outpui-- Included a Standard-3 & t4 din Dot matrix. 1-PLaserlet and PostScript