engineering - worldradiohistory.com...wave- analysis-part 2 a simple electronic switch electron...

ectroricEngineering

INCORPORATING ELECTRONICS. TELEVISION AND SHORT WAVE WORLDAO.

PRINCIPAL

CONTENTS

Beryllium Copper-A New Alloy

Wave- Analysis-Part 2

A Simple Electronic Switch

Electron Optics-(Electronics Group Lecture)

The Cathode Follower-Data Sheets Nos. 39, 40 and 41

PDF compression, OCR, web optimization using a watermarked evaluation copy of CVISION PDFCompressor

fii ElectroniclEngineering December, 1942

For Radio and Telecommunication, X-ray H.T. Rectifiers, Relay and Contactoroperation, Electrical Precipitation equipment, Surge Absorbers, Metallic Sputterin gapparatus, H.T. and E.H.T. Insulation Test equipment, Instruments and Power

applications, etc.

PEEL WORKS, SALFORD, 3. Telephones: BLAckfriars 6688 (6 lines). Telegrams and Cables : "Sparkless, Manchester"

PROPRIETORS: THE GENERAL ELECTRIC Co. Ltd., OF ENGLAND

The fact that goods made of raw materials In short supply owing to war conditions are advertised Inthis magazine should not be taken as an indication that they are necessarily available for export.


'December, 1942 Electronic Engineering 271

13

MINISTRY OF SUPPLY

NON-FERROUS METALS

ARE URGENTLY NEEDEDUpon them largely depend the accuracy, deadliness and efficiency of Britain'sinstruments of war. Practically all of them have to be imported across thou-sands of miles of dangerous seas, making heavy demands on shipping space.

Make a clean sweep of your unwanted

COPPER, ZINC, LEAD, PEWTER, WHITE METAL

BRASS AND BRONZE, ALUMINIUM SCRAP

The need for vast and increasing quantities of these metals must be met by a nationalsearch for everything made from them which is broken or serving no useful purpose.

WHAT TO SEARCH FORHere is a list of some of the important non-ferrous metal articles,parts and equipment likely to be found on your premises. Useit as a guide to your search for everything you can spare.COPPER - cable, electrical parts and fittings, sheathing, tube,

turnings, wire.ZINC - sheet, cuttings.LEAD - covered cable, pipe, sheet, solder.WHITE . METAL - anti -friction metal, plumbers jointings,

solder waste.BRASS - pipe, sheet, tube, turnings.BRONZE - bearings, bushes, cocks, couplings, crown wheels,

junction boxes, unions, valves.ALUMINIUM SCRAP (and its alloys) - pipe, sheet, castings,

tube, turnings.

HOW TO DISPOSE OF IT1 Sell your non-ferrous scrap to a Merchant.

2 Or hand it in to a Local Authority Depot.

3 SPECIAL COLLECTIONS of amounts over ONE TONmay be obtained by getting in touch with the nearest Demolitionand Recovery Officer. If you don't know his name, write toThe Ministry of Works & Planning, Lambeth Bridge House,London, S.E.I.

NOTE : Under the provisions of the Scrap Metal (No. 2) Order,1942, if you are in possession of more than 3 tons of Scrap Metal,it is now an offence not to disclose the fact to The Ministry of Works& Planning, Lambeth Bridge House, London,

WANTED! INDUSTRIAL SALVAGE STEWARDSSalvage is of such great and increasing importance that it should be made the personal responsibilityof one particular individual in the various departments of every organisation. Appoint your own

Industrial Salvage Stewards. Put them .in sole charge of an intensive drive for

STILL MORE PAPER, RUBBER, RAGS, BONES, KITCHEN WASTE-AND METAL


277 Electronic Engineering December, 1942.

PRE-EMINENT IN PEACE

INDISPE A IN WAR

ADVERTISEMENT OF THE TELEGRAPH CONDENSER CO., LTD.GP 5273


December, 1942 Electronic Engineering 273

Preparation of MetallurgicalSpecimens for the MicroscopePurity and correct grade of materialsare of paramount importance in themanufacture ofWestinghouse Metal RectifiersOne of the preliminary processes isthe examination of crystal formationby means of the microscope, forwhich the specimen is first groundand then polished, to provide anoptically flat surface,

With nearly all our output ear-marked, we regret that Westing-

house Rectifiers are so difficult to obtain. Butbehind the ever increasing demands, technicianswork ceaselessly after new developments. Afterthe war there will be available to industryWestinghouse Rectifiers of remarkably advanceddesign-born of the persistent research nowgoing on.

GUS

WESTINGHOUSE BRAKE &SIGNAL CO., LTD.

PEW HILL HOUSE,CHIPPEN HAM, WILTS.

RESISTANCE WIRESTO THE FINEST SIZES

MOLYBDENUM

VAC -STEEL

PLATINUM SHEATHED

GAS -FREE NICKEL

WIRES, TAPES, BARS.MESH & VALVE PARTS

8, Vape4

VACTITE WIRE CO. (1919), LTD.19, QUEEN ANNE'S GATE, WESTMINSTER, S.W.I

TELEPHONE : WHITEHALL 2552

ElectronicEngineering

DECEMBER, 1942

Volume XV. No. 178.

CONTENTSPAGE

Editorial 275

Beryllium Copper - A New Alloy 276

Wave Analysis Part 2 280

The Manufacture of Accumulators 283

A Simple Electronic Switch ... 284

The Testing of Magnetic Materials by theC.R.O. 286

The Cathode Follower Data Sheets Nos. 39,40 and 41 ... .. 287

Electron Optics ---Electronics Group Lecture 295

Frequency Mixing with a Triode-Hexode 300

Modern Design 303

Taking Care of Transmitters 304

Abstracts of Electronic Literature 306

Patents Record 308

Notices of Meetings 308

A Pulse Generator for Circuit Testing 310

CONDITIONS OF SALE-This periodical is sold subject to thefollowing conditions, namely, that it shall not without thewritten consent of the publishers first given, be lent, re -sold,hired out or otherwise disposed of by way of Trade except at thefull retail price of 21- and that it shall not be lent, re -sold, hiredout or otherwise disposed of in a mutilated condition or in anyunauthorised cover by way of Trade, or affixed to or as part ofany publication or advertising, literary or pictorial matter

whatsoever.


274 Electronic Engineering December, 1942

Valves and ViaductsTo the engineer, accurate knowledge of

the presence and precise form of vibrationsis a matter of paramount importance.

The structural engineer needs to defineand evaluate the stresses and strains onmassive girders and framework; the mechani-cal engineer is concerned with the smoothtransmission of power and the elimination ofheavy wear and tear on plant and machinery.

With the aid of Mullard Valves inspecially designed apparatus, vibrationshaving an amplitude as low as a few micronscan be easily and accurately measured.

In almost every branch of engineeringpractice, problems are constantly being solvedand the field of knowledge widened throughthe application of the thermionic valve - andMullard have the right valve for the purpose.

MULLARDTHE MASTER VALVE

A Valve for Every PurposeDOMESTIC COMMERCIAL INDUSTRIAL SCIENTIFIC MEDICAL EXPERIMENTALTHE MULLARD WIRELESS SERVICE COMPANY LIMITED, CENTURY HOUSE, SHAFTESBURY AVENUE, LONDON, W.C.2. (44)



PROPRIETORS :

HULTON PRESS LTD.,Electronic

EngineeringEDITOR

G. PARR

EDITORIAL, ADVERTISING AND PUBLISHING OFFICES, 43-44, SHOE LANE, LONDON, E.C.4

TELEPHONE:

CENTRAL 7400

Monthly (published last day of preced-ing month) 2/- net. Subscription Rates :Post Paid to any part of the World -3 months, 6/6; 6 months, 13/- ; 12

months, 26/-. Registered for Trans-mission by Canadian Magazine Post.

TELEGRAMS :

HULTONPRES LUDLONDON.

ROBERT BOYLE, Who has beencalled the father of modernchemistry, was not above making

a bit of a mystery when it suited hispurpose.

In one of his lesser knownpamphlets, published anonymouslyin 1678 and re -published later underhis own name, he describesan experiment in metallurgicalchemistry which is interesting torecall in these days of new andwonderful alloys.

The pamphlet is written in theform of an account of a discoursegiven by BOYLE to a company offellow chemists (who frequentlyinterrupted him) and is verbose andtedious in parts.

Briefly, it tells of his meeting witha mysterious Stranger from theEast (they always are !) who gavehim a screw of paper containing arare Powder, together with Adviceon how to experiment with it.

BOYLE was " both surprized andtroubled to find so very littlePowder " on opening the paper, anddid not dare to weigh it in case helost it on the balance. However,having got the necessary witnesses-a wise precaution in those days ofalchemy and fraud-he proceeded tomelt 2 drachms of gold -in a crucibleand cast the Powder in, allowing it

" An account of a Degradation of Gold, made byan Anti -Elixir : A Strange Chemical Nyrative.'(Herringman, London.)

Acknowledgment is made to Cdr. R. T. Gould,whose book " Enigmas " furnished the reference.

C

Alchemyto " defuse " throughout the gold.

The results, when cool, astonishedhim as the gold became convertedinto a brittle alloy " looking likeBrass or worse " with totallydifferent properties from the originalmetal.

This is where our opinion ofBOYLE as an experimenter is lowered.In reply to a question from a scepticin the audience he had to confessthat he did not test the alloy withnitric acid as he had none in stock.

ELECTRONIC ENGINEERINGMONOGRAPHS.

The Publishers of this Journal arearranging to issue at intervals a seriesof Technical Monographs on speciallyinteresting radio and electronic subjects.

They hope that the production ofsuch monographs will enable radioengineers and students to build upat reasonable cost a library of specialisedinformation which could only be otherwiseobtained from a number of books andperiodicals.

The monographs will be in the formof octavo size booklets of uniform style,bound in stiff red paper.

The first has already appeared underthe title

FREQUENCY MODULATIONby K. R. Sturley, Ph.D., A.M.I.E.E.

based on the articles which have recentlyappeared in this Journal, revised bythe author.

Copies can be obtained from technicalbooksellers, price 2/6, or direct fromthe Publishers at the above address,price 2/8 including postage.

His account of the subsequent testsis vague, although he did manage tofind the density, which wasapproximately 15 (gold is 19).

After hinting at other still morewonderful properties of the Powder,which he was not at libertyto disclose, he concludes " ThisExperiment plainly`shews that Gold,though confessedly the most homo-geneous and least mutable of metals,may be chang'd both as to malleable-ness, colour, homogeneity andSpecifick Gravity "-and by a" despicable quantity of Matter."

On another page in this issue,Dr. HUNT, with less mystery,describes how copper can be changedboth as to malleableness, colour, andother characteristics usually as-sociated with it, by an almostdespicable quantity of beryllium.

Beryllium was first isolated in1838, or was it ?

If the mysterious Stranger hadonly given BOYLE a little more of thewonderful Powder . . .

At the conclusion of the meetingin 1678, one of the chemists utteredwhat might fairly be called amouthful :

" We ought not to be so forwardas many men of otherwise greatparts are wont to be in prescribingnarrow limits to the power ofNature or Art, and in deriding thosewho pretend to believe uncommonthings in Chymistry as Cheats orCredulous."



Beryllium CopperA High Strength Alloy for Electrical and Instrument Springs

By L. B. HUNT, Ph.D., A.R.C.S., M.Sc.*

THE development of electronicengineering has presented manyproblems to the metallurgist;

who has been prevailed upon duringthe past twenty years or so to providea number of alloys having specialcombinations of electrical andphysical properties. The majority ofthese newer materials have been re-quired for use inside the radio valveor the cathode-ray tube, but in othercases special properties have beenfound necessary or advantageous inthe purely electrical parts of thecircuit.

One of the principal requirementsof designers in this industry is a re-liable material for use as current -carrying springs, possessing the opti-mum combination of elastic strengthand electrical conductivity. Manysuch applications arise in radioengineering, chiefly in componentssuch as wave change switches, volumecontrols, appliance plugs, micro -switches, fuse clips, vibrators andmeasuring instruments. For moder-ately light mechanical duty the useof phosphor bronze ornickel silver forthese parts has become standard practice, but in many cases springs inthese materials are found to be de-ficient in either mechanical or elec-trical properties, or sometimes in both.* Mallory Metallurgical Products, Ltd.

In this event, beryllium -copper isfound to give the necessary combina-tion of high elastic and fatiguestrengths with high conductivity. Aswill be seen from the detailed proper-ties of this alloy given in Table I,beryllium -copper offers an outstand-ing balance of electrical and mechan-ical properties which make it suitablefor springs designed for very severeduty.

Other non-ferrous spring materials,such as' phosphor -bronze and nickel-sil ver, acquire their strength andelastic properties by severe cold work,but the provision of spring propertiesin this way gives rise to difficultieswhen further working is necessary informing the spring from strip or wire.Where the design of such parts is re-latively simple, hard rolled strip can,of course, be used, but in many casesmore or less severe drawing or form-ing operations are involved and onlyannealed or lightly cold rolled mate-rials are permissible, with the obvioussacrifice in spring properties. Further,the electrical properties of thesealloys are by no means high, normallybeing of the order of to to 55 per cent.that of copper. Beryllium -copper notonly offers greatly superior elastic andfatigue strengths, but also improvedconductivity and the great advantageof being readily fabricated in the soft

state and then heat treated to give therequired spring properties. Springs orother parts of complicated shape maythus. be produced in beryllium -copperwhich would be incapable of produc-tion in a hard rolled material, and canthen be given a tensile strength -up to90 or 95 tons per square inch, with alimit of proportionality of 5o tons persquare inch.

Composition and Properties

Table I gives a comparison of theproperties of beryllium -copper withthe average properties of extra -hardphosphor bronze (6 per cent. tin, 0.2per cent, phosphorus) and high carbonspring steel in strip form. It will beseen that the spring properties ofberyllium -copper are very little in-ferior to those of high carbon steel andgreatly superior to those of extra -hardphosphor bronze, a material which isnot capable of any great degree ofdrawing or forming. (The Erichsenvalue of this temper of phosphorbronze strip will be approximately 3.0mm.). Beryllium -copper not, ofcourse, magnetic, and has the furtheradvantages of excellent resistance tocorrosion and wear. The high limitof proportionality of beryllium -copperpermits of very much higher safestresses than may be employed whenusing phosphor' bronze, while its

Fig. I. Typical beryllium -copper components of electrical and pressure measuring instruments, snap action switchesand radio apparatus.



higher endurance or fatigue limit alsoenables considerably greater workingstresses to be employed in springs sub-ject to rapid and continuous deflec-tion. Thus a beryllium -copper spr'ingcan be of smaller cross section thanone of phosphor bronze, without lossof current -carrying capacity, and itslength can be decreased to obtain thedesired deflection. For these reasonsthis alloy is being employed to agreatly increased extent for parts ofelectrical equipment and instrumentswhen no other material can providethe combination of workability withhigh mechanical properties in thefinished product.

Beryllium -copper has been knownto metallurgists and engineers forsome years now, but has not alwaysbeen as readily available as was de-sirable. Normally the alloy has con-sisted of 2.25 to 2.5 beryllium, withthe balance copper, but more recentlyit has been found that the addition ofa small percentage of cobalt givessomewhat improved and more uni-form physical properties. Such analloy has been made available in thiscountry during the past two years andit is with this material, known asMallory 73 beryllium -copper, that thisarticle is concerned. The nominalcomposition of Mallory 73 is beryl-lium 2.0 per cent., cobalt 0.5 per cent.and copper the remainder. It isavailable in all those forms likely tobe required by electrical, radio, andinstrument manufacturers, includingstrip down to o.002 inch thick, wiredown to o.00i inch in diameter, hair-spring strip down to o.oio inch byo.0005 inch, round and oval tubing,and contact bi-metal strip having asilver inlay or facing.

The use of this material for elec-trical and instrument springs hasother subsidiary advantages besidesthose already outlined. For example,in the final heat treatment of thesprings or other parts all internalstresses are relieved, with the resultthat the possibility of " season crack-ing " does not arise, while undesir-able effects due to creep, drift orelastic hysteresis are also reduced tovery small proportion. Again, a singlespring in beryllium -copper can re-place an assembly of steel backingspring and current -conducting lead.Where slightly elevated temperaturesare involved in service, beryllium -copper has the further advantage thatit may be used up to 2oo° C. withoutloss of strength. The modulus of elas-ticity of beryllium -copper, approxi-mately 18.5 x so' lb. per square inch,is slightly higher than that of phos-phor bronze, but very much lower thanthe figure for steel. Springs in thisalloy, therefore, have considerable

Fig. 3. An assembly of beryllium -copper contact blades, having an electrodepositedcoating of silver and rhodium, employed in radio apparatus manufactured by Murphy

Radio Limited.

amplitude of movement within thevery great elastic range.

Production of SpringsNo difficulty is experienced in the

manufacture of spring parts in beryl-lium -copper provided that due care istaken in the selection of the most ap-propriate temper of strip or wire. Itshould be borne in mind, however,that beryllium -copper work -hardensrather more , rapidly than the usualbrasses and bronzes. For this reasona softer temper should be used thanwould normally..be chosen in the caseof phosphor bronze, for example.

Mallory 73 Beryllium -Copper hasbeen made available in four tempersof sheet or strip, the quarter -hard,half -hard and hard tempers being ob-tained by imparting increasing reduc-tions in thickness, while wire is simi-larly available in three tempers. Theproperties associated with each ofthese tempers are set out in Tables IIand III. The properties given hereare, of course, those of the alloybefore final heat treatment. For deepdrawing purposes, for bending flatover very small radii and where no" spring -back " is allowable in form-ing, beryllium -copper strip should be-employed in the soft or fully an-nealed state. The quarter -hard tempershould in general be employed forlight drawing operations and for worksuch as severe bending over a com-paratively small radius. For moresimple forming operations the half -hard temper should be used, while thefull hard condition is preferable forthe blanking of fiat springs or partsinvolving only light bending or set-ting operations. In the case ofMallory 73 beryllium -copper in wireforin similar calditions obtain. The

annealed temper should be employedfor very severe bending and similaroperations, while the quarter hard andhalf hard tempers will be found moresuitable for progressively less severeforming. For coil springs wire in thehalf hard condition is normally mostsatisfactory.

In any given case beryllium -coppershould be used in the hardest temperwhich will satisfactorily form theparts, as the final strength and hard-ness obtained on heat treatment will -be slightly higher for increasedamounts of cold work, while the pos-sibility of distortion occurring duringheat treatment is greatly reduced withthe harder tempers.

The manufacture of hair springs forinstruments is carried out in exactlythe same manner as for phosphorbronze. After winding in the usualbarrel or' pill -box," springs aregiven their heat treatment as des-cribed below. This serves not only toset the spring, but also to develop therequired spring properties. It will beseen, therefore, that while othermaterials normally suffer some reduc-tion of elastic limit during the settingoperation, beryllium -copper actuallyincreases very markedly in strength.

No attempt should be made to carryout bending or setting operations ofany kind on beryllium -copper afterfinal heat treatment on account of itsvery high elastic limit and springproperties.

Heat TreatmentThe exceptional physical properties

of beryllium -copper are obtained, asalready mentioned, by a simple low -

temperature heat treatment. This pro-cess is a function of time and tempera-4,0re, hardening occurring more



400

350

300

250

200

150

100

315°C

300°C

(275 °C

PRECIPITATION

-ANNEALED

FROM

CONDITION

HARDENED_

3 4HOURS

5

400

350

300

250

200

150

100

300°C315°C

275°C

III

PRECIPITATION HARDENED_

COLD ROLLED

FROM

CONDITION

D 2 3 4 -

HOURS

Fig. 2. The effect of time and temperature on the hardness of Mallory 73 beryllium -copper developed during heattreatment.

rapidly at higher temperatures. Ifheating is continued for too long, orif the temperature is too high, somere -softening is brought about, moreparticularly in hard rolled strip orhard drawn wire., Material in the an-nealed or lightly worked condition isless susceptible to this effect but, onthe other hand, develops slightlylower properties and requires a longertime and a. rather higher temperaturefor optimum hardening. These pointsare illustrated in the two sets ofcurves. In the left hand curves, forannealed material, it may be seen thatat 275° C. hardening takes place veryslowly. At 300° C. hardening occursmore rapidly, but considerable time isrequired to develop maximum proper-ties. At a higher temperature, how-ever, such as 315° C., hardening ismore rapid and satisfactory resultscan be obtained at this temperatureafter a treatment lasting 2 hours.

The curves on the right hand side,for strip rolled half -hard, show first,that appreciably higher hardnessfigures can be obtained by comparisonwith annealed material, and secondlythat this is brought about at ratherlower temperatures and with shortertimes. At 275° C. hardening is stilltoo slow to be convenient, but it ispossible to hold the alloy at this.tern-perature for a considerable time with-

out re -softening. At 300° C. harden-ing occurs fairly rapidly and a reason-able latitude is possible in the time forwhich the work may be held at tem-perature. At 315° C., however, a veryrapid hardening is followed by somere -softening, and the permissible mar-gin in the time factor may be foundto be too narrow for satisfactory com-mercial heat treatment. Best resultsare thus obtained with the followingtreatments :

From annealed condition, 2 hours at310 to 320° C.

From quarter hard or half hardcondition, ri hours at 30o to310° C.

From hard condition, i hour 'at 300to 310° C.

This treatment may be carried outin an 'air furnace provided that theparts are suitable for subsequentcleaning by pickling. If this is notthe case, as with very small ordelicate springs for example, the useof a hydrogen or nitrogen atmospherefurnace is necessary. Care should betaken in heat treatment that thecharge is actually at the temperatureshown on the pyrometer, the hot junc-tion of the thermo-couple being placedas close as possible to the work. Thespecified period for heat treatmentshould be taken only from the timewhen the charge reaches the proper

temperature. Cooling may be carriedout either by quenching or in air.

To avoid distortion during heattreatment, finished springs and simi-lar parts should be adequately sup-ported. Flat parts may be clampedbetween heavy steel plates, whileformed parts should be clamped inappropriate jigs, or wired together inpacks if their design makes this pos-sible. Small parts may alternatively belaid in trays and surrounded withsand. The risk of distortion is muchless in the harder tempers.

The times and temperatures givenabove represent good average practicefor the development of maximumhardness and spring properties, but incertain cases or for special require-ments it may be necessary to modifythese treatments to some extent. Ad-vice in this connexion will readily begiven by the manufacturers of beryl-lium -copper, who will also undertake'the heat treatment of finished parts forusers who are not themselves equippedwith suitable furn tces.

Soldering and _BrazingBeryllium -copper may be soft sol-

dered without difficulty after finalheat treatment. It may also be silverbrazed successfully, but in this casethe jointing operation must be carriedout before- heat treatment.



TABLE IProperties of Beryllium -Copper compared with Phosphor Bronze and

Spring Steel

Beryllium-Copper

PhosphorBronze

SpringSteel

Ultimate stress, tons/sq. in. ... ... 80-90 45-50 95-100

Limit of proportionality, tons/sq. in. ... 47-50 28-33 60-65

Brinell hardness ... ... ... ... 350-420 185-225 ' 400-450

Fatigue limit, tons/sq. in. ... ... ... ± 20-22 ± 10-12 ± 35-40

Modulus of elasticity, lb./sq. in., x 106 ... 18-19 15-17 29-30

Electrical conductivity, per cent. 1.A.C.S. 25-30 12-15 3-4

TABLE IITempers of Mallory 73 Beryllium -Copper Strip

(Properties before final Heat Treatment)

TemperVickersHardness

Tensile Strengthtons per sq. in.

Elongationper cent.

Erichsen valueMillimetres

Annealed ... 100-120 30-35 45-60 8.0-9.0

Quarter Hard ... 160-185 35-40 25-35 5.5-6.5

Half Hard ... 200-225 40-50 10-20 4.0-5.0

Hard ... ... 230-250 50-55 5-10 2.5-3.0

TABLE IIITempers of Mallory 73 Beryllium -Copper Wire

(Properties before final Heat Treatment)

TemperTensile Strength

tons per sq. in.Elongation per cent.

on 2 in.

Annealed ... ... 35-40 25-40

Quarter Hard ... 40-50 5-10

Half Hard ... ... 50-60 2-5

When parts to be soft soldered canbe washed after soldering, it is recom-mended that a zinc chloride flux beused, and in these cases no difficultyis found in making joints, using eithera soldering bit or a torch. If thematerial has not been pickled, the sur-faces to be joined should be cleanedwith fine emery. After. the joint ismade the assembly must bethoroughly washed to remove alltraces of flux and prevent subsequentcorrosion. Many articles, includingthe majority of electrical components,cannot be washed after soldering,while it is frequently stipulated inelectrical work that no flint other thanresin may be used. In these casesgood joints can be made with cer-tainty, using resin as a flux, providedthat sufficient care is taken to cleanthe surfaces immediately before sol-dering. When parts are in an inac-cessible position for cleaning, the sur-faces to be. joined should be tinnedbefore assembly. Resin cored soldermay be used, in which case no addi-tional flux need be applied. Success-ful results have also been obtained by

painting with a solution of resin inmethylated spirits. Once the surfaceshave been tinned they may readily bejoined on any later occasion, using aresin cored solder.

When high -strength joints are re-quired beryllium -copper may be suc-cessfully silver brazed, but this mustbe done before final heat treatment asthe temperatures involved are, ofcourse, much higher than the harden-ing temperatures. A low meltingpoint silver brazing alloy such asEasy-flo (melting point 63o° C.)should be used. Mechanical cleaningbefore silvei brazing is not essential,and equally good results may be ob-tained on oxidised as ,on freshlypickled material. The only degreas-ing treatment required is a pre-liminary warming with the blowpipeflame before applying the Easy-floflux, which is easily able to cope withthe oxide film formed on. beryllium -copper. The alloy behaves very simi-larly to brass in brazing and the normal technique may be employed usinggas -air or oxy-gas heating. Thenormal heat treatment should, of

course, be carried out after brazing.It will be seen that beryllium -cop-

per again offers a marked advantageby comparison with other non-ferrousalloys in that strong and effectivejoints can be made, or electrical con-tacts fitted, by silver soldering, with-out the final mechanical propertiesbeing thereby affected. This pro-cedure is not, of course, possible onany alloy relying upon cold work forits strength and hardness, as the tem-perature involved would be more thansufficient to anneal the material..

Advantages in Electronic EngineeringIt will be appreciated from the fore-

going that electrical, radio and instru-ment manufacturers now have avail-able for special purposes a materialpossessing greater strength and elasticproperties than any other non-ferrousalloy, a conductivity higher thanthose of the brasses and bronzes,and at the same time possessingexcellent workability. As comparedwith the normal type of springmaterial which relies upon coldwork for its elastic properties, in thecase of beryllium -copper a simple heattreatment develops these properties,with a consequent simplification ofmanufacturing and assembly prob-lems.

Again the possibility of makingjoints or attaching electrical contactsby silver brazing is of considerable im-portance in electrical engineering,while the availability of silver or sil-ver alloy faced bi-metal strip opens upstill further possibilities in designingcontact arrangements for strength andsimplicity.

For components required to carryradio -frequency currents, beryllium -copper may also be given an electro-deposif of silver, or of silver plusrhodium. In this way a heavy dutyspring is provided, with the silver tocarry the R.F. current and therhodium to protect the silver fromtarnish and to maintain a low, con-stant contact resistance. Rhodium andsilver electrodeposits are also beingemployed on beryllium -copper to givea wear resisting contact surface inwave change switches subject toparticularly severe duty.

These advantages of beryllium -cop-per have rapidly been recognised byradio engineers, and its use has be-come standardised for a variety ofcontact blades and clips, special relaysprings, micro -switch blades, fuseclips and similar parts. Unfortunately,many of the interesting applicationsof this unique alloy cannot be des-cribed in detail until they are oncemore directed towards human comfortand enjoyment.



Wave AnalysisPart II - Analysis of Semi -Periodic Wave -Forms

By K. BOURNE*IN a previous article (Part I),t

wave -forms were classified underfour broad headings. Those waves

in Class D, i.e., " transient," arenormally treated only from atheoretical point of view, and there islittle of practical value to be gainedby attempting to analyse these elec-trically. The wave -forms in Class C,i.e., continuous but non -recurrent, areof great importance in acoustics, in-cluding as they do the human voiceand musical instrument tones, singleor in combination. When the wave-form is continuously changing, theanalysis must obviously be made asquickly as possible if it is to give aclose approximation to the instantane-ous spectrum of the wave; however,many musical tones and sustainedvowel sounds are sufficiently steady inpitch and free of incidental noise thatthey may be considered as Class A orB waves. Even in this case, speed isvery desirable; fortunately, a verymoderate degree of accuracy is suffi-cient, since in most cases the tones,under investigation cannot be exactlyreproduced, and an approximatespectrum gives enough information re-garding the character of the sound.

It is convenient to note here afundamental division in the methodsof analysis of these waves : betweenmethods which utilise the tone at thetime of its production, and thosewhich record it by some graphicalmethod, and analyse the recordmechanically or optically. The firstgroup has obvious advantages in dis-posing of the necessity for recordingapparatus, and in giving a direct andrapid result; on the other hand, thegraphical record of a wave -form (andin practice this means a record onsound -film, variable -area or variabledensity) gives a permanent result, cor-rect at any and every instant to thelimits of accuracy of the recording ap-paratus (and these limits can be madevery close indeed). Thus the film re-cording can be made to yield preciseinformation as to the value of the har-monic components of a. wave -form,giving frequency, phase and ampli-tude : but the extraction of this in-formation is not easy., and the filmmethod totally fails when the spec-trum is required to be instantly vis-ible (as in following continuouschanges of frequency or quality of atone). ..

* Post Office Research Stationt Electronic Engineering, Sept. 1942,13. 249.

Methods using Film RecordingsWhilst various methods can be de-

vised for the continuous graphical re-cording of electrical wave -forms, thecommercial types of recordings onphotographic film, or on transparenttape with an opaque coating (e.g., thePhilips -Miller recorder) have beenbrought to a very satisfactory degreeof accuracy over the audio -frequencyband, and are the only methods likelyto be used at present. If it were de-sired to record high -frequency wave-forms for exact analysis a precisecathode-ray technique would appearthe best method, or possibly the use ofa Kerr cell.

There are three main methods ofanalysis of film recordings : opticaldiffraction methods, mechanical orgraphical analysis, or electro-mechanical means.

In optics, the diffraction grating is,in its simplest form, a series of linesruled very closely and regularly upona transmitting or reflecting surface.When the ruled surface is illuminatedby a beam of light, the -spaces betweenthe lines act each as a source of light,and the wave -fronts from each sourcecombine beyond the grating to givealternative maxima or minima oflight, according to the phases of thesuperimposed waves. The positions ofthe maxima are dependent upon thewavelength of the light, and thus thegrating will 'give the spectrum of theincident light. The regularity of thespectrum given is dependent upon theaccuracy of the grating, and muchcare is ,taken to ensure its absoluteregularity. If, now, a monochromaticlight source is used, but falls upon anirregular grating, the dispersion pro-duced will be a function of the grat-*ing, and indicate the nature of thegrating irregularity. It was stated byBrown' that the film record of a sound,when used as an optical grating,would give an immediate indication ofthe nature of the sound. The theoryhas been worked out in detail bySchouten,' who shows that for a filmrecorded on the variable intensity sys-tem, only the first order of the spec-trum is present, and this spectrumconsists of lines representing the tonespresent in the recording, spaced on alinear frequency scale, their intensi-ties being proportional to the squaresof the amplitudes of the individualtones. If the recording is variableamplitude, the diffraction pattern isvery much more complicated (for a

complete treatment, see Schouten,Physica, 7,2, February, 1940), but onthe horizontal axis of the pattern, thespectrum of the sound is representedas with the variable intensity film.

This method gives a visiblespectrum for any sound, even a singleimpulse, and is thus well adapted toindicate the nature of a waveform ina direct and rapid way : where con-siderable accuracy is required, thereare evident optical limitations, moreparticularly in the measurement of theamplitudes of the individual com-ponents.

A more accurate 'analysis can becarried out by enlarging the filmrecord (which must be amplitude re-corded) to a suitable size, and analys-ing the curve obtained by one of theschedule methods for Fourieranalysis' or by the use of a speciallydesigned form of planimeter (the oneusually employed is the HenriciHarmonic Analyser).' By either ofthese means, accurate information onthe amplitude and phase of all har-monic components present can be ob-tained; but if many components arepresent, both methods are verylaborious and slow.

There are two electro-mechanicalmethods which greatly decrease thework involved in the analysis of acurve. One fairly obvious method' isto take the strip of film on which thesound is recorded, or an enlargedcopy, and fasten it around a circularwheel so that the ends abut with aslittle discontinuity as possible, andtransform the curve back into elec-trical vibrations by an orthodox sound -head. This backward procedure willgive a steady output of harmonics ofthe period of rotation of the circularfilm; if the length of film has beencarefully chosen to correspond withthe fundamental period of the recordedtone (or a multiple of this period),these harmonics will give the spectrum,of the recorded tone. The electricaloutput can be analysed at leisure byany of the methods developed for theanalysis of steady periodic wave-forms.

The most rapid method yet devisedappears to be that of Montgomery,'who claims an analysing time of about112, minutes. This method assumes aperiodic function, but since it employsonly a single period, it gives a closeapproximation to the instantaneousfrequency spectrum of a changingwave -form. The film recording (which



may be variable density or variableamplitude) is combined in an opticalsystem with a variable density filmbearing the recording of a pure sinewave; by using a series of films bear-ing different frequencies, the succes-sive harmonic components of therecorded sound can be obtained.

The recorded wave -form can be ex-pressed as the following Fourier Series

co

f(x)=ao+ ao cos nx + ft,. sin nxa a

co

= co +1c. cos (nx-8.,)1

where au=c. cos 0,,bo= co sin 6,,

and to these series apply the well-known formulae

I27f

a = - (x) cos nx. dx.o

2.77-

= - f(x) sin nx. dx.o

zirc. = - f(x) cos (nx - 0.)dx

o

where2.71-

f(x) sin (nx - dx = 0

2'77'

ao = co = - f(x) dx27r o

The optical transmission of the filmrecording of f(x) is A + f(x) (where Ais large enough to make A + f(x)always positive), and of the secondfilm, B[I + M. cos (nx - 8)] whereM. is somewhat less than z, B. isaverage transmission of screen, and 0gives position of second film relativeto first (which is assumed to be asingle period of (x) occupying alength of 277- units). When, by theoptical system, the two films aresuperimposed the transmission isgiven by

2'77'

T ...-fB.[A+1(x)][i +M. cos (nx-8)] dxO

Simplifying and substituting fromequations above.T =277-Bo(A+ co)+,7rBoMoco cos .(8 - 9,,)

If now 9 is varied from 0 to 7, thedifference in the values of T is217-BoMoco cos ea= viTB.M.a. ;and simi-larly from 0 = '712 to 0 = 37/2, thedifference in T is 271B.M.c. sin 9,, =27B,,M,,b.. Thus from a knowledge ofthe constants of the sine wave record-ing, the harmonic components of therecorded sound can be found, by usingsuccessive values of n. The amplitudeof the nth harmonic is actually c =Vas' + b'; this can be found directly,

for let T be a maximum ate = 0 anda minimum at 0 = 00 + IT; then thedifference is 27B./11,,c,,.

LightSource.. 11,1_

Al

PhotoCell

o>

Fig. I. Optical arrangement of Montgomery'smethod.

The optical system used to facili-tate the superposition of the films isshown in Fig. z. The film bearing thewave -form 1(x) which is to beanalysed is placed at A, and an en-larged image at B is adjusted so thatone period just fills a window of pre-determined size. The successive filmsof cos nx are brought up and slidacross behind this window : the varia-tions of photo -cell current are ampli-fied and recorded on a chart by meansof a level recorder. The apparatus asdescribed in Montgomery's articlegives a fundamental frequency rangeof 65 c/s-3io c/s (using the customaryrecording speed for sound film), anddeals with harmonics up to the 3oth,thus the total range covered is ex-tended to 9,3oo c/s : the smallest har-monic amplitude measureable is about2 per cent.

Methods not using recordingIf useful results are to be obtained

on short duration and changing wave-forms, speed of analysis is essential.There is an inevitable compromise in-volved here between speed and resolv-ing power, since the more selective afrequency -discriminating device ismade, the longer its build-up time.Those methods which use a singleresonating element are obviously at afurther disadvantage, since the fre-quency range must then be covered

sufficiently slowly to allow the elementto build up at any point where a com-ponent may be present. Thus themore rapid methods use a multiplicityof selective elements, the limit in thisbeing reached when a diffracting grat-ing is used. Other elements used areband -filters and vibrating seeds.

The band -filter method has beenadopted by Freystedt,' who uses aseries of 22 filters, 3 for each octave,to cover a range of 4o c/s to 5,500 c/s.The general arrangement is shown inFig. 2; the outputs of the band -filtersare connected successively to a cathoderay oscillograph so that the amplitudeof the sound averaged over each bandis shown as a line on the oscillograph.The time of build-up of a band -filter isapproximately equal to 11(12 - fi)where f, and f, are the cut-off fre-quencies' so in this case for the lowestfrequency filter the build-up time is ofthe order of z/zo second. The timetaken to cover the frequency rangewith the switching mechanism is thusconveniently made z/io second, asthere is no point in making it faster. Itis seen that any attempt to improve theresolution will result in proportion-ately slower response : in any case,there are practical limits to the num-ber of band -filters which can beemployed.

A very similar method' has beenused to obtain statistical data onmusical instruments and orchestras,finding the peak and average power inthe bands covered : this is, however,hardly wave -form analysis in thestrict use of the term:

Greater resolution at the expense ofbuild-up time is obtained in the assem-bly of reeds which has been called anAcoustic Spectrometer." In this instru-ment, 48 reeds per octave are used.The frequencies of_ the reeds arechosen to increase in geometric pro-gression : their response (which is ap-

c

ti

2

3 /Set of 22 I

Band -Filters I

ti

Modulator

1

22/GangedSwitches

- II Ii -

r\J3000

ti

J LCathode-rayIndicator

Fig. 2. Freystedt's band-pass filter method.



Fig. 3. Response of reeds in Hickman's apparatus.

proximately constant for all reeds) isshown in Fig. 3. This corresponds toa logarithmic decrement 8 of .o072,giving a rate of decay of f/8 db. persecond, where f is the resonant fre-quency of the reed concerned; thus areed at 130 c/s (a low speech fre-quency) cannot follow changes quickerthan some 16 db/sec. : this speed issufficient for most speech and muchmusic. Another limitation is that the

amplitude at resonanceratio (which

amplitude at zero frequencyequals gr/28) is 46.7 db. and so theamplitude of a reed due to a com-ponent of its frequency cannot begreater than 46.7 db. above that ofany lower frequency reed in responseto the same tone. This low selectivityfor frequencies some distance from theresonant point is, of course, inherentid any selector consisting of only oneresonant element for each frequency :a tuned circuit before the reed wouldeliminate the trouble, which is, how-ever, of little consequence in thenormal uses of the instrument.

The wave -form to be analysed isnormally a sound received by a micro-phone and amplified in the normalmanner before being fed to the reeds.The reeds are driven so that theirsecond mode of vibration (6.3 timesthe first) is eliminated (by driving onthe node for this vibration); the thirdmode is 17 times the first, and thosereeds with frequencies less than 1/17of the top frequency to be analysed arespecially shaped to remove this mode.The general arrangement of the reedsis shown in Fig. 4. The light fallingupon the concave mirrors on the reedtips is focused on to a screen, andlines of length proportional to thereed amplitudes give a visible spec-trum of the sound.

The extreme limit in speed has beenreached by Meyer," who performs theanalysis by a diffraction grating. Thegeneral arrangement is shown inF ig. 5 ; the sound is received by a highquality condenser microphone, afriplifled and modulated up to a frequencyat which a grating of reasonable sizecan be used. The frequency bandchosen is 45-5o kc/s. (for an audioband of o-5 kc/s) and cannot be muchhigher, since the absorption of the airis becoming quite high. There is alsothe problem of radiator and receiver;the sound source is a very thin duralu-

min ribbon,. 5 A thick by 6 mm. by120 mm. in a field of 20,000 gauss,through which the output of the H.F.amplifier passes, whilst the receiver isa small' condenser microphone (dia-phragm io mm. diameter of ioduralumin foil). The acoustic grat-ing consists of 300 .steel needles of 3.4mm. diameter fastened betweenparallel iron plates 12 cm. apart,forming a concave grating 3 metres inlength. The elementa7ry theory of adiffraction grating is applied byMeyer to obtain the resolving power,which in terms of frequency dis-crimination is f In where f is the fre-quency, n the number of grating ele-ments. Thus in the case where f isabout 45 Kc/s and n = 300 the resolv-ing power is 15o c/s. It is shown byMeyer that a simple concave gratinggives large aberrations, and a con-siderable amount of empirical modifi-cation to the grating curve is neces-sary to obtain good results. The finalgrating has a resolving power of 125c/s. and a dispersion of 8 cm. peri,000 c/s.

The output from the ribbon loud-speaker falling on the grating is re-flected by the steel rods and the inter-ference pattern thus produced is cal-culated so as to spread the componentfrequencies over a curve in front ofthe grating. The condenser micro-phone moves along this path, and theamplified output operates a recordingoscillograph; the beam of light fromthe oscillograph is moved at right -angles to the recording direction insynchronism with the movement of themicrophone. Thus a spectrum is ob-tained on a frequency base, andis recorded photographically; photo-graphic considerations limit thespeed to about i/io second. Thisis not the ultimate speed, whichis given by the time required to buildup the complete diffraction pattern,and is -actually the difference in timetaken by the extreme sound raysto reach the grating and reflect back tothe microphone position. Ii accord-,

Fig. 4. Layout of Hickman's apparatus.

ante with the general rule thatresponse time is inversely proportionalto resolution, this time is somewhatunder ihoo second. Although givingsufficient resolution over most of theband, this method is not altogethersatisfactory in that secondary maximaappear after each main component,the first of which has an amplitude 20per cent. of the principal amplitude:it is evident this may be very confus-ing in a crowded spectrum.

BibliographyI Brown, Nature 140, 1099, (1937)

see also Germansky, Annalen derPhysik 7.453 (5934

2 Schouten, Philips Technical Review, 3,298, (1938) : see also Schouten, -Philips Technical Review, 4, 290 (1939)for illustrations of results obtained.

-3 See, for instance, Grover, Bull. BureauStandards 9, 567 (1913) and Taylor,Phys. Rev. 6. 303, (1915).

4 Henrici, Phil. Mag. July 1894.Miller, Jour. Franklin Inst. 185, 285,(1916).Encyclopaedia Britannica ioth Ed.Article Mathematical Instruments."

I Sacia, Jour. Optical Soc. of Am. 9,487, (1924)

6 Montgomery, Bell. System Tech. Journ.17, 3, 406 (1938) (U.S. Patent No.2,098,326).

7 Freystedt, Zeit. fur. Tech. Physik, 16,533 (1935)

8 Kupfmuller, E.N.T. 1, 546, (1924) andE.N.T. 5, January, 1928.

° Sivian, Dunn and White, Jour. Acoust.Soc. America, 2, 3, 33o (1931).Hickman,10 Acoust. Soc. America,6,Meyer, Jour. Acoust. Soc. America, 7,88 (1935). This is a translation byF. A. Firestone, of the original German.

CondenserMicrophone

&dancedModulator

11

H.PFilter

4510

toPhotographic

OscillographPlate /inked to Microphone

Ribbon /L S ////

\\

\\Moving

Microphone \\\ \I11's'4..

Fig. 5. Schematic diagram of Meyer's apparatus.



D

The Manufacture of Accumulators

Casting lead grids in pairs for the battery plates,

Joining batches of plates to form the battery andmelting on the terminal post.

Stages in the manufacture of radio and carbatteries illustrated by photographs takenat the works of the Edison Swan Electric

Co., Ltd.

Spreading paste of active material on thelead grids.

(Above) Fitting wood separators between plates.(Left) Sealing in tops of finished batteries with

compound.

These photographs were taken from a short instructional film :

" The Acid Test," distributed by Wallace Productions,- Ltd.Copies of the film are available for Service instructional purposesand enquiries should be addressed to the Publicity Dept., TheEdison Swan Electric CO., Lcd., 13a, King Street, Twickenham,

.Middlesex.



VALUES OF COMPONENTSR1 = I megohm R, = 15,000 ohms

R, - 1 megohm R, = 40,000 ohmsR, = 40,000 ohms R, = 56 ohmsR, = 40,000 ohms R10 = 56 ohmsR5 - I megohm R11 = 56 ohmsR, = I megohm R12 =

R13 =0.2 megohm1,500 ohms

V1 = V2 = MULLARD ECH35

ClC2C3C,C3Cl'C,C8

= 500 iip.F= 500 pl.&= 0.5 tIF= 0.5 µF= 0.5 I.LF

= 32 pf= 16 pf= 50 µF

Fig. 5. Complete circuit diagram of the electronic switch using two triode hexodes.The photograph on the right shows the assembly on a'chassis measuring61 inches square.

WITHIN the last few years,number of electronic switchcircuits have been devised

with the object of extending the scopeof the cathode-ray oscillograph, bymaking it possible to view simultane-ously two or more phenomena whenusing a single beam tube. Each cir-cuit had some particular merit, but ingeneral the more successful requireda considerable number of valves,while the more economical usuallygave a bad " haze " between the twotraces._

The' principle underlying the elec-tronic switch is very simple. The twovoltages E, and E2 to be studied areapplied to the valves V, and V, whichhave a common anode load RA. Bysome means, V, and V, are renderedalternately conducting at a frequencyzn cycles/sec. The output voltage E.across RA will then be represented byFig. .2 a , and when this is applied tothe vertical deflector plates of acathode-ray tube, the two voltages willappear as shown in Fig. 2b. If thesteady anode currents la, and 4,2 ofthe two valves had been the same, the

two traces would have appeared on acommon base line, so that variation ofthe steady anode currents provides aconvenient way of shifting one tracerelative to the other for identification,etc.

Another point is readily apparentfrom Fig. 2a. If the changeover fromV, to V, is not instantaneous there willbe a transition stage when both valveswill draw current resulting in a sud-den drop in output voltage E0. Theseso called transition voltages are thecause of the background haze fre-quent'y associated with electronicswitches.

It is most convenient to use a pairof multi -grid valves fOr V, and V,' sothat the switching may be effected bya suitable voltage applied to theauxiliary grid, thus ensuring thatthere is no interaction b tween signaland switching voltages. In order toreduce the transition time to a mini-mum, the auxiliary grid used for con-trol should have a short base, and asharp mit-off, while it is advantageousthat it should lose control of the elec-tron stream when it is more than a

A SimpleElectronic

SwitchBy A. W. RUSSELL

(Mullard Wireless Service Co.)

F. I. Diagram illustrating the principle of theelectronic switch.

few volts positive with respect tocathode, as otherwise the waveform ofthe switching voltage will be super-imposed on the output voltage. Con-sideration of these points leads to thechoice of the hexode as the most suit-able valve, since grid 3 is situated be-tween' two screening grids maintainedat a positive potential of ioo volts, asshown in Fig. 3a. The char-acteristic of the hexode portion of aMullard .ECH35 triode-hexode isshown in Fig. 3b.



The generation of a suitable switch-ing voltage is the next consideration.It is fairly obvious from the foregoingthat the ideal waveform would besquare topped with " crossovers" assharp as possible. Such a waveform,or a close approximation to it, isgiven by the well known multi -vibrator circuit of Fig. 4. The fre-quency of vibration n of such an oscil-lator is given by the formula ts= i/4RCwhere R is measured in megohms andC in Microfarads. This circuit using

/ triodes, following on the choice. of apair of hexodes- for the switchedvalves, leads one irresistibly to at-tempt to satisfy all the conditionsusing a pair of triode-hexodes only.Since in the triode-hexode the grid ofthe triode portion is connected inter-nally to grid 3 of the hexode, no ex-ternal coupling is required and thewhole circuit becomes very simple.

Fig. 5 shows a practical circuitwhich has been evolved for use withthe Mullard Oscillograph GM.3156.The frequency of oscillation of themulti -vibrator has been fixed at 500

VA H 300VG 2 100VGI a -3

+ G 004v

G2+100V

20

15

I0

5

3B

-30 -20 -10 +10VG 3

Fg. 3. Connexions and characteristics of ECH 35triode hexode.

cycles/sec. for this particular work,but can be varied by changing thevalues of Ci, C2, R, and R2. Thechoice of the switching frequency isgoverned by two consideration's.When both input circuits are earthed,the output voltage will be a squarewave and this must be handled with-out distortion by the oscillographamplifier. If the frequency of the

a

Eci

Esr

0A

0Yiv

T

Fig. 2. (a) Wave form of output voltage acrossanode resistance of Fig. I and (b) the waveform on

the screen of the tube.

square wave be N cycles/sec., thenin order to preserve the flatness of thetops of the waves, the amplifier musthave a response flat to N lo sinusoidalcycles/sec., while in order to handlethe transient " crossovers," it mustalso be flat up to ION sinusoidalcycles/sec. The amplifier of theGM.3t56 is flat from o.s cycles/sec. toio,000 cycles/sec. and can thereforehandle square waves with frequenciesfrom s cycle/sec. to i,000 cycles/sec.

There is only one point in the finaldesign which calls for special com-ment, and that is the method of ob-taining bias for the hexodes and fosshifting one trace relative to the Other.This method was chosen for two rea-sons. In the first place it preventscross -modulation of the two inputvoltages due to a common cathode im-pedance and in the second place itprevents unwanted coupling betweentriode and hexode portions of thevalves.

b

Fig. 6a shows the wave form of theswitching voltage generated by themulti -vibrator, while Figs. 6b and Gcshow the electronic switch in opera-tion : The two.traces show the ampli-tude and phase relationships betweenthe grid and anode voltages of a back -coupled i,000 c/s. oscillator. Fig. 6bwas obtained with a low value of grinleak only just sufficient to maintainoscillation, .while in Fig. 6c it was in-creased to a much higher value, giv-ing rise to distortion of the waveform.

The experimental model described,together with a power pack to permitoperation direct from the 50 cyclemains, can be assembled on a boxchassis measuring only 61 in. by 6a in.

Fig. 4. Basic multivibrator circuit.In conclusion, it may be helpful

briefly to consider the best way ofusing the electronic switch. Theoscillograms Figs. 6a and 6b were ob-tained when' the electronic switch wasfollowed by an amplifier having again of approximately 30, giving asensitivity of approximately 30o mV.rms/cm deflection. Taking into ac-count the amplication of the elec-tronic switch, this gives an overallsensitivity of approximately 20 mV.rms/cm. The limitation on the ampli-fication which can usefully follow theswitch is set by the magnitude of theswitching voltages and the squarenessof the multivibrator waveform. Withthe switch described, quite good re-sults can still be obtained with a gainof too, but as a general rule, if theinput voltages to the switch are lessthan too mV. it is best to use a stageof amplification in front of the hexodesrather than increase the gain follow-ing the switch.

Fig. 6. (a) Waveform of switching voltage (b) and (c) typical waveforms obtained from the switching circuit.



The Testing of Magnetic Materials by the C.R.O.

ar0

CAL

IsmEAs

L2

Specimen

00 42.30is ,Co 9s

sobs

1 Meg

CRO

1/vfeg.

Complete testing circuit for magnetic materials.

IN the September issue of the il our-nal of Scientific Instruments,* K.Kreielsheimer, of the New Zealand

Dept. of Scientific and IndustrialResearch describes a method of ob-taining the B -H curves of magneticmaterials using an electrostatic de-flection C.R. tube. If a voltage e, =E. sin- cot is supplied to the primaryof a transformer having no lossesand no secondary load, the alter-nating current traversing theprimary coil gives rise to an alter-nating field and flux practicallyin phase with this current. Accord-ing to Lenz's law an e.m.f. will be in-duced by this changing flux in theprimary and secondary winding inequal phase, but the primary e.m.f. isof opposite sign to the supply voltage.Hence, the primary e.m.f.E, = - E. sin wt = - to -8 NiqdB Idt,where N. = number of 'primary turnsand q the cross-section of the core. Itfollows :B = -' (E.IN,q) cos wt x to" and E2 =.

E.(N.IN.) sin wt.This simple ahalysis shows clearly

that all voltages available on thetransformer terminals are sine func-tions, whilst the induction B is acosine function, i.e., goo out of phasewith either primary or secondarye.m.f., and that the voltage amplitudesare proportional to B.. It is then ob-vious that a correct hysteresis looputilising the voltage drop across a p. 137. Vol. 29. 1942.

small resistor in the primary circuit asproportional to H and the primary orsecondary voltage as proportional toB can only be obtained if the voltagewhich is' to represent the B -deflectionhas undergone a phase shift of 1-7.

We know from consideration of theB -H loop that B... and H... are cor-responding points. Hence, the maxi-mal deflection in the horizontal shouldcoincide with the greatest vertical de-flection. This co -phase condition,however, no longer holds near thecentre f the loop where remanenceand coercive force indicate a phasedisplacement between H and B. AsB follows a cosine law, H or -the mag-netising current im in the primarymust show a considerable distortionfrom the cosine shape. It is, therefore,not permissible'to adjust the range offlux densities over which the speci-men is to be investigated -by means ofa variable resistor in the primary. Be-cause of the distorted form of theprimary current the voltage dropacross this resistor would cause theinput voltage to the transformer to .beno longer a sine curve. It follows thatthe resistance inserted in the primarycircuit to create a voltage drop pro-portional to H must be small com-pared with the input impedance of thespecimen transformer. Any variationof B... required is best obtained by acontinuously variable supply trans-former (e.g., a Variac).

The importance of their phaseshift' of the output voltage has beenrecognised by Johnson' and Cosens2who load the secondary coil of a trans-former with a condenser and a resist-ance in series. By making R>t/wCutilising the voltage across C for ver-tical deflection, a very close approachto a goo phase shift is obtainable, butonly. by sacrificing most of the avail-able output voltage. It might bepointed out that the above considera-tions are not influenced by the actualpresence of losses. The primary cur-rent i, is composed of the wattlesscomponent, i., the magnetising cur-rent, and the loss component, i.e.,which determines the area of thehysteresis loop. With a negligiblesecondary load, the terminal voltageof the secondary coil equals E, whichwhen advanced go° will coincide withthe phase of i..

This necessary rotation of the volt-age vector by can be achieved sim-ply and accurately by using a phase -adjusting device, previously des-cribed.' Although in this arrangementthe secondary is loaded by a resistor.and condenser in series the funda-mental difference from Johnson andfront Cosens lies in the fact that R isadjusted to be equal to r/wC. As hasbeen shown in the publication re-ferred to, the voltage' between themidpoint. or artificial midpoint, ofthe secondary winding and thecommon point of R and C isthen goo out of phase to E. and ofhalf its -value. Fig. r gives the com-plete circuit diagram'proposed for thedetermination of the iron losses. Asindicated, various R -C combinationscan easily be provided to enable theinvestigation to be carried out at vari-ous frequencies f, for which R=1/0C.When selecting the values of the R -Ccombination it should be kept in mindthat the load created should benegligible.

The method used was found to bevery sensitive and results obtained ap-pear to be accurate within a per cent.Routine tests could easily be ar-arranged as a comparison method witha standard specimen excited by thesame magnetising current, whilst thehysteresis loops could be superim-posed on the oscillograph screen bymeans of an electronic switch.

I JOHNSON,(1929).

2 COSENS, C. R. Wireless Eng. /2, p. 190 (1935).3 KRE/SISHEIMER, K. Wireless Eng. 17, P. 439

(1940).

J..B. Bell System Tech. J.-8, p. 286


SUPPLEMENTARY TO DATA SHEETS 'Lb

The Cathode FollowerBy C. E. LOCKHART

THE "cathode follower " type ofcircuit has the distinction ofbeing one of the most versatile

circuits used by the Radio Engineer.The circuit in its most elementaryform is shown in Fig. i, and derivesits name from the fact that an inputsignal that makes the grid go positivewith respect to H.T. negative willalso make the cathode go positivewith respect to H.T. negative. Inaddition if the resistance Re is madesufficiently high the output voltage ispractically equal to the input voltage,and the cathode may therefore be saidto follow the grid as far 'as voltagefluctuations are concerned.

The main advantages of thei` Cathode Follower " type of circuitmay be summarised as :-

I. High input impedance.2. Low output impedance.3. High anode impedance.4. Freedom from distortion.The circuit shown in Fig. 1 repre-

senting the most elementary form ofnegative feed -back,. has been treatedon a number of occasions,' ..a' but asfar as the author knows a generalisedtreatment of the "Cathode Follower"circuit has not been published before.

Owing to the complelity of thegeneralised treatment it is proposed toseparate the problem into two stages ;the first part limits the problem to lowand medium frequencies where inter -electrode capacity displacement cur-rents are negligible compared withthe electron currents and electron in-ertia effects can be neglected. Thesecond part treats the problem in e.

more generalised manner essential atthe higher frequencies.

General RelationsIf we assume linear valve charac-

teristics we can express the anode cur-rent of a triode by

V. + g

1.- (I)R.

where V. and V, are the static poten-tials applied between the anode andcathode and grid and cathode respect-ively. I. is the steady anode current,and R. is (d1.1dV.) or the Anode A.C.resistance of the triode. (For a list ofsymbols see Data Sheet).

For an input signal of " E,1 " volts(Fig. za) the signal " E." appliedbetween grid and cathode is (E,-i.Z)and the change in anode -cathode voltsis (- i.Z) where i. is the resultingchange in anode current and 'Z is theimpedance in the cathode circuit.

= i+gRc Eo

TBgRc_

Fig. I. (a) Simplest form of CathodeFollower circuit (b) & (c) Its equivalentcircuits (for 11. >> I) from which the vol-tages E0 across any load connected betweenterminals A & B can be calculated. Thecapacity C is assumed to be an infinite by-pass condenser.

£g --

Ei gRc' 1+9,8c+paCcRe

C

A

Ep

AA". LA

EM

=TRekkoCcite

+g +' TB

Fig. 2. (a) Cathode Follower circuit withstray capacity Cc across the cathode resistorRc (b) & (c) Equivalent circuit when µ >> Ia nd when inter -electrode capacities areneglected.

Therefore

- (2)R.

and subtracting (i) from '(2)14E,

i. -R. + Z(i +

gEi(3)

gZ(i + ilpThe output voltage E. is (i.Z) and thestage gain E./E, is given by

gZVoltage Gain = (4)

+ gZ(i + i/P)The grid -cathode voltage E,=EI-E.or from (i) and (2)

+µE5 µEi. -

thereforeR R. + Z

(5)

gZ I14

Eg = E, (6),Z(I +

The output impedance of the deviceis obtained by calculating the currenttaken from a voltage source E. (Fig.za) when E1 is zero. This current" i. " is

E. E. + µE.i.=

Z R.Therefore the admittance ilo=i.!E. is

ZA. = -+ g

\14)

and the output impedance is approxi-mately equal to the impedance Z inparallel with a.tesistance /g when pyj I. The equivalent circuits areshown in Figs. rb, 2b, and c.

Due to the negative, feed -back pro-duced by the cathode load the anodeA.C. resistance for voltage fluctuationsbetween anode and H.T. negative isincreased to R.'. From (2)

v. - i.Z(1 + µ)i. -

thereforev.

R.'=i.

R.

(7)

=R.ri +g2(i+ 001 (8) -

Cathode LoadThe most usual load employed in

the cathode is a resistance " Re"(Fig. za) and at low frequencies Z inthe above expressions may be re-placed by " Re" (see Fig. 1). It is,however, impossible to eliminatecapacity shunt across the cathodeload, due to say the anode -cathodecapacity of the valve, stray wiringcapacities and the capacity of any fol-


egg1

lowing stage. As the trequency of theinput signal is increased so the effectof " C. " becomes increasingly impor-tant. The impedance of Re and C. inparallel is given by

R.Z -

j6;Co(Re +Re Re

(9)+ jcoReC e I+ jPo

if we let co. = r/R.C. then wReCe == Po.

The -value of Z given by (9) can nowbe inserted in the expression pre-viously obtained.

Anode -CurrentInserting the value of Z from (9) into(3) we get

g(i +1. - Ei

(1 + gRe + R.elRo + jP0)rationalising

(r+g Re+RelRo+P.2).-jpo(gRei. - Ei.g

gRe- Et

(i+RelRo+gRe)'+po2where

/9 (Ib)

P..8 = tan -1 (19)

+ R./R.+ gR.)when µ

.gRe iE0=-Ei 19

1+ gReq 1+[(25.1(1 + gRe)f(20)

where 0 = tan-' (Po ) (21)1+gRe

The phase, angle 0 is plotted againstpo for different values of gRe on DataSheet No. 0.

Voltage GainAs the voltage gain is given by the

ratio EolE, all the expressions for E,

+RelRo)

(i + gRe + RelRo) + poa

if we now express i, in the formEilgKI

V +gRe+ RelRo+ p.2>+ to2(gR+12.11?.)'io = Ei.g

(i + gRe + Re/R.)' + po'where

po(gRe +- tan -1

+ gRe + RelRo + pog(12

It should be noted that the term(i +gRe+R.IR.)=1+gR.(x+ i/µ)

so that the relation ReIR. <gRe issynonymous with it> I.

Grid -Cathode VoltageFrom (6) and (9)

(I +RelRo)+jPoE. = EI

(i+RelRo+gRe)+jp.(13)

(io)

Fig. 4. Vector relations of cathode followerstage at higher values of 2/r/CeRc=Po

V I_ + R./Z.) (x+ Rol R.+ gRe)+p.92 +(pogRe)'=Et Ich

(i +RelRo+gRe)'+Po'pogRe

where 0=tan-' [(i +Re/R.) (1+Re/R.+gRe)+P.11

Output VoltagegRe

E. _ El(i+RelRo+gRe)+jPo

rationalising

[1 +R./R. +gR.)- jP.]E0=E, gRe (17)

(r+ReIR.+gR.)'+po2

(16)

(14)

(15)

Ea

Fig. 3. Vector diagram of cathode followerstage for 211C,R, =Po 0

in the above paragraph become ex-pressions of voltage gain whendivided by Ei.

Relative GainThe term " relative gain " M is

used to express the ratio of the gainat a frequency f to the gain at a verylow frequency, i.e., /5. 0.The gain at very low frequencies isgiven from (15) by

gRegain at P.=o- /o° (22)

I + Ref/Z.+ gReso "that the relative gain M in db. is :M in db

20 log i P.I +

i+RIRo+gR

or when RIR. <i or 14>

M In db = 20 lOgioV

(23)

1+[pogi + g R e)12(24)

This expression has been plotted onData Sheet No. 39 for values of po upto io and gR. products up to 20.

Time DelayWhen considering video amplifiers

in television we are concerned notonly in the gain response character-istic, but also in the linearity of thephase angle 8 versus frequency inorder that signals of different fre-quencies shall take an equal time topass through the amplifier (see DataSheets and XXIII-XXV).

The time difference between the out-put voltage vector and input signalvoltage vector reaching their maxi-mum amplitudes for a signal of fre-quency f is given by

t = - (25)2121

In the case of the cathode followercircuit Fig. 2.0 is always negative andtherefore t is negative. The outputsignal is thus time delayed withrespect to the input signal. Also as

= o when f = o the value of t from(25) will give the time delay of theoutput voltage vector of frequency frelative to that of the output voltagevector of very low frequency whenthe input signals of these two fre-quencies start simultaneously.Now from (19) and (25)

tan-'( po

i+Re/R.+gR.)ti - secs. (26)

27rf

In order to enable a generalisedgraphical representation to be made


the expression (27) must be rewrittenin the form

P.tan-'(

+ Re/R.+gle)foti- secs. (28)

2/rPo

which simplifies to

tan -1(+gke

foti= secs. ( 29)

27250

when A> 1. Expression (29) has beenplotted on Data Sheet No. 40.

Vector RelationsThe problem is better visualised by

a graphical illustration of the vectorrelations expressing the equations de-rived above. Taking first the case ofvery low frequencies or po= (aCR:zothe vector relations for Fig. 2a areillustrated in Fig. 3.

Here we have for the referencevector the input signal E,; the anodecurrent i. and the output voltage E.are in phase with Ei as both andare zero. The angle 95 is also zero sothat Eg=(Ei-E.) is in phase with El.

If we now take the case of a con-siderably higher frequency inputsignal, while still maintaining E, asthe reference rector (see Fig. 4), thenthe anode current vector will now leadthe input signal E, by the angle V,given by equation (12). The outputvoltage vector E. will, however, lagbehind E,' by an angle 0 as Z iscapacitive. The vector for Eg, beingthe difference between the vectors Eiand E. will lead Ei by an angle q.

A very significant feature which 122-' comes at once apparent, is the pos-sible rapid rise in the grid -cathodevoltage Eg when E. is no longer inphase with El. Equation (i4) whichsimplifies to :

V [(1 + gRo)+ p.92 + (PogRo)2Eg=Ei (30)

(i + gRe)' +when Re/R.< i and µ3j I must there-fore be used to check that the signalEg does not drive the valve into gridcurrent or off its characteristic at the

highest frequency to be employed.Input Admittance

In the relations so far derived thecapacity current through the cathodeimpedance Z due to the grid tocathode capacity Cs. (see Fig. 2) hasbeen assumed negligible comparedwith the anode current. We can pro-ceed to determine the input admit-tance of the stage with the above as-sumptions and provided the electroninertia effects are small enough toensure negligible grid loading.

The Input Admittance A, is given bythe ratio of the current taken from the

source of the- signal E, to the voltageEl. As the grid loading is assumednegligible this current is equal to thecurrent through the capacities Cs.and Cs., therefore

?c.go +.4 =

Eiand i g. = Ei (j Cg,)

=cg. = Eg(iwC sc) (32)From equations (3!) (32) and (13) wehave

(1 + RoIR.+j25.)Ai=j63Cgo + jwCg.

(i+gRo+RoIR.)+ip.(33)

Rationalising we have-PogRo+j((i+RoIR.)(1+gRo+ReIR.)+P.9

/it = jo.)Cs. + coCs,(i + gRo+ReIR.)2 + p.2

From which the input resistance Ri is(I + gRo+ RoIR.)2 + poi

R, =.Q.CgoPogRo

(1+ gRo)' +pozwhen II> 1 (36)

and the input capacity isCi=Cg.+

((I +R./ R.) (1+ gRo+RoIR.)+31.1Cgo

(i+ gRe+RoIR.)2+po2+ gRo+P.

Cgo (38)(i+gR0)2+p02

when 12./R. < z and p.>Cgo

I -FgRo(39)and Ci Cg. + ,

when in addition po is small comparedto unity.'

From the complete solution given inPart II it will be seen that the aboveequations are accurate at low values ofpo provided gRo> CgolCo + Re/R,)and at higher values of po if in addi-tion Cse/C. < t.

The interesting facts derived fromthe above expressions are that thegrid -cathode capacity Cs. is reducedby the factor (1 + gRe) and that theinput resistance of a cathode -followercircuit as shown in Fig. 2 is negative.The latter fact could have been predicted from the vector diagram Fig.4, as the current iose must be repre-sented by a vector leading the vectorEg by 909. If this current vector isnow resolved into two components onein phase-quadrature (imaginary com-ponent) and the other in phase (realcomponent) with the vector Ei, it willbe seen that the real component isnegative in sign.

If we consider the case of a normalamplifier with a. resistive load R. inthe anode circuit, then in the case of avalve with an anode A.C. resistance

(35)

Re(41)

2.8

R. large compared with Re, the volt-age amplification is given by :

Gain = gReUsing this gain as a reference figure,

we can consider that the effect of con-necting the resistance in the cathodecircuit (Fig. i) is to reduce the gain by(I + gRe) and to reduce the grid -cathode capacity by the same ratio.

Output ImpedanceFrom Eq. (7) we have :

I ZZ. = -

A. i+gZ(i + 19)Re

1- -E gRe(1+11A)+jP.

(34)

(40)

VEI+gRo(1+11,1)r +Pf

where 7 =tan' [i+gRo(1+11µ)

(42)

With high slope valves gRe may bemade, large compared with unity, andfhen at low frequencies expression(4i) reduces to the output impedance,being i/g. Thus with a mutual con-ductance of io mA/V. an, ottput re-sistance of only ,00 ohms is obtained.

Anode A.C. Resistance

The anode A.G. resistance R.' of thecathode follower is by (8) :

gReR.' = R.[1 +

I+iP.For any voltage fluctuation v. be-tween anode and H.T. negative, thefraction of v. developed across Re isgiven by :A.C. volts across output

+ 01)1 (43)

A.C. volts across H.T. line

=v. R./R.RoIRa+ gRo(1-F ilp)+jP.

(44)with an absolute value of

R.

R. I/ [i + RoIR.+gRo(i +11µ)]2+P.2.(45)

When po is J.ery small, such as atlow frequencies, but with it> i andgRo> i, expression (44) simplifies to :

A.C. volts across output v.= -

A.C. volts across H.T. lineThe complete solution of the cathode

follower circuit together with generalcircuit considerations, will be given inthe next series of Data Sheets.


29cDATA SHEETS XXXIX, XL, AND XLI

The Performance of the Cathode Follower CircuitTHE use of these Data Sheets is

best illustrated by working anexample. Suppose it is desired

to investigate the performance as acathode follower at video frequenciesof a triode -connected pentode having amutual conductance of g = 5 mA/V,ti = 8a, and a total stray capacityacross the load of 7o µµF.

Let us first calculate the perform-ance with a cathode load of 2,000ohms.

(1) The value of C.R. = 0.14 x io-',and therefore

tonfo = - 1.14 Mc/s.

217- x 0.14Also we have : gR0 = to, and R. =plg = 16,000 ohms.Therefore Re/R.= 1/8.The gain for po = o is 101(1 + io) =0.91.

(2) Relative Gain.With the above constants and the

gR0 = to curve of Data Sheet 39, wecan now plot the curve of relativegain versus frequency. For this pur-pose we first express the applied fre-quency in terms of po (=f1f.). In theattached table the values of f havebeen tabulated in the first column,while po(=f1f.) is given in the secondcolumn.

From the values elf po it is now pos-sible to read the relative gain M fromData Sheet 39. The values so ob-tained are given in the third columnof the table.(3) Phase Angle.

In a similar manner the phase angleof the output voltage E. relative tothe input voltage E1 may be read offData Sheet 41.

The values so obtained are given incolumn 4 of the table.(4) Values of Time Delay.

To obtain the values of the TimeDelay we can either use the ex -

LIST OF SYMBOLS

Ai = Input Admittance of cathode follower.A. = Output Admittance of cathode follower.

Cac, Cga, Cgc = Anode -cathode, grid -anode, and grid -cathode inter -electrode capacities,respectively.

Cc = Total' effective capacity across cathode load resistance R0.Ci = Input Capacity.Ei = Input signal voltage between grid and negative H.T.E. = Output signal voltage.Eg = Signal voltage developed between grid and cathode.

The above voltages are assumed to be of sinusoidal form and must all be expressedon the same basis, i.e., in R.M.S. or Peak volts. The instantaneous value of the voltageis of the general form e = E sinwt= E sin 2oft.

= Frequency of applied signal.

, i.e., frequency in c/s at which co. Cc Re = 2-rfo2ir Cc Re

g = Mutual conductance of a triode = dia/dvg.is = Signal component of the anode current.

E.M = Relative amplification in db. = 20 logia-.

Ei

Po = 0, Co Re = flfoRa = Anode A.C. resistance of a triode = dia/dva.R'a = Anode A.C. resistance of cathode follower.Re = Resistance of cathode load in ohms.Ri = Input resistance of cathode follower.t = Time in seconds.

= Time delay of frequency f relative to zero frequency.Phase angle in radians Phase angle in degrees

2 77 f 360.f

f=fo =

va -z =zi =zo

=tP

=0=Y

A.C. voltage applied between anode and H.T. negative.Impedance of cathode load circuit.Input impedance of cathode follower.Output impedance of cathode follower.Amplification factor = gRa.Phase angle of is relative to Ei.Phase angle of Eg relative to E1.Phase angle of Ea relative to El.Phase angle of output impedance of cathode follower.

Cc Rc =

pression :Bo

t, = secs.360o.f

where 8 is given its appropriate nega-tive sign, or read off (fot,) directlyfrom Data Sheet 4o and then obtainti by dividing by f. = 140,00o c/s. Thevalues of fat, have been tabulated inthe fifth column and the time delay(-ti) expressed in microseconds in the

TABLE

last column. For very low frequencies(i.e., po 0).

lot, - secs.277(1 + gRe)

C.R.and to - secs.

g/2The calculation of time delay for

the case of a modulated carrier willbe dealt with in part II.

AppliedFrequencyf in Mc/s.

p° = f/f.Relative

Gain M indb.

PhaseAngle

0°fa t1

Time Delayin iLS

ti).

0 0 0 0 0.0145 0.01271.0 0.88 -0.06 4.5 0.0144 0.01262.0 1.76 -0.11 9.0 0.0143 0.01253.0 2.64 -0.25 13.4 0.0141 0.012354.0 3.52 -0.42 17.7 0.014 0.012255.0 4.4 -0.65 21.8 0.0138 0.01216.0 5.28 -0.9 25.5 0.0136 0.01197.0 6.15 -1.18 29.2 0.0135 0.0118


er

4

U LC)

CC

Qc°

(-)

0k cN

t

k4)

W 0 0

40

f\I


23

A A

OT

r"

V) a_

r

(f)

''ocO

0

- n -



Battle of WitsIT takes experience and a good head to ride the production

problems of today. By good fortune the emergency is bringinginto the saddle brains and ability equal to the task.

Certainly the Simmonds Organization-youthful, un-conventional, dynamic-irresistibly attracts brilliant men andwomen, but the proof of the pudding lies in the many parts playedby the products of the Simmonds Organization in widely diversefields of essential production. There each day they are successfullyintroducing new methods to quicken the national pace.

SIMAIONDSTHE SIMMONDS NUTSIMMONDS-CORSEY CONTROLSSIMMONDS CONTENTS GAUGESFOR AIRCRAFT.SIMMONDS PQSITION INDICATORS

SIMMONDS ELECTRIC TEMPERATUREAND PRESSURE GAUGES

SIMMONDS AIRCRAFT FLOORING

SPIRE"..".V.e" NUTSFRAM OIL E ENGINE CLEANERSIMMONDS INDUSTRIAL ANDMARINE CONTENTS GAUGESSIMMONDS JOINTING COMPOUNDSIMMONDS CRYSTAL UNITSSIMMONDS-GOUDIMENAVIGATIONAL INSTRUMENTS

SIMMONDS AEROCESSORIES LIDA COMPANY OF THE SIMMOND S GROUP

LONDON MELBOURNE PARIS NEW YORK.

ALL ENQUIRIES TO GREAT WEST ROAD, LONDON.



Electron OpticsA lecture delivered before the Electronics Group and the Midlands Branch

of the Institute of Physics on the 31st of October, 1942

By D. GABOR, Dr.Ing.*NTRODUCTION. Electron opticswas born fifteen years ago, when

Hamiltonian methods had been used. we are talking about, and establishesMerely the terminology was borrowed a natural connexion with the older

H. Busch proved in an important from optics, but the authors, almost part of electron optics. These newtheoretical paper that the magnetic without exception built up their re- dynamical devices, such as the veloc-field of a cylindrical coil acts as an sults from the consideration of elec- ity modulation tube, the Klystron, orelectron lens. Such coils, known as tron trajectories. This situation is the Betatron have no counterpart in" concentrating coils " had already well reflected in the numerous excel- light optics, but they are fullybeen used by a full generation of lent treatises on electron optics. The covered by the Hamiltonian Analogy;physicists, without anybody noticing Hamiltonian analogy adorns a few if it is suitably extended to four in -their lens effect. Busch's paper was pages, for the rest it is just dynamics stead of three dimensions..more than an eye opener; it was of electrons in electric and magnetic The Scope of Electron Optics. --Italmost like a spark in an explosive fields, translated finally-when pos- appears now necessary to re -define themixture. In 1927 the situation in sible-ihto optical terms. This is not scope of electron optics. A limit mustphysics was such, that nothing more surprising, seeing that the computers be drawn somewhere, or else the sub -than the words " electron lens " were of optical instruments do not use ject would cover the whole theory ofneeded to start a real burst of creative Hamiltonian methods either, but de- electronic devices. It is not sufficientactivity. On the experimental side velop their lenses by " ray tracing," either to say that electron optics ex -the work of Richardson, Wehnelt, which is closely analogous to the com- tends as far as the HamiltonianLangmuir and others had provided putation of electron trajectories. Analogy, - as some of the `° transiteasily controllable sources of slow Until three years ago I. took pre- time " devices which we want to in -electrons, and the production of high cisely the same view of the practical dude in the subject are not coveredvacua was a matter of everyday value of ;he Hamiltonian methods as by the analogy in the restricted senselaboratory routine. On the theoretical most other workers in the field. I in which it is often used. But if wesides two years before. Busch's paper looked often with respect and admira- adopted this narrower interpretation,Schrodinger discovered wave me- tion at their beautiful exposition in we should have to exclude also spacechanics, by combining certain ideas of Born's " Optik," or their adaptation charges4 If, however, we allowde Broglie with the half forgotten by Walter Glaser to electromagnetic space charges, i.e., the action of-elec-analogy of mechanics and optics fields, but it never occurred to me to trons on one another inside the

outer circuits shouldwhich Sir William Rowan Hamilton use the methods. For the last three vacuum device,had worked out almast exactly a hun- years I was not an active worker in claimed that the outer

might be logically

dred years earlier. By the time Busch electron optics, and dismissed the be also included. Indeed, the mathe-.had established the analogy of axially subject entirely from my mind. But matician could treat these by thesymmetrical electro-magnetic fields when I had the pleasure of accepting' same multidimensional Hamiltonianwith lenses by somewhat laborious the invitation to give this lecture, I methods. As an entirely logical divi-calculation of trajectories, the found that I must have done some un- sion appears impo5sible, I proposeHamiltonian analogy was again so conscious mental digestion in the with the inevitable measure of arbi-familiar to theoretical physicists, that meantime Suddenly I .discovered, trariness, to consider electron optics

astheir immediate reaction was :-" Of that certain problems on which I had co -extensive with electron dyna-course, we could have told you so ! laboured in vain with trajectories, mics, including -the theory of spaceWe could have told you even more, could be solved almost effortlessly charges ,and space currents, but ex -There are not only electron lenses, with Hamiltonian methods.t As I eluding their .reaction on the outerthere must be a whole optics of elec- want my experience to benefit others, circuits.trons. You have only to look it up I intend in this lecture to adhere more The Ballistic Problems as Illustrationin Hamilton's writings and trans- closely to Hamiltonian ideas than of the Hamiltonian Analogy.late it into your experimental usual. Instead of dealing in generalities, I

prefer to explain the optical treatmentlanguage !" I see also another justification for of mechanical problems in the simpleIlut the experimental physicists this. In the last six or seven years example of the ballistic problem. Thedid not bother much about the electron . optics has definitely out- Newtonian treatment of this problemHamiltonian methods. The words grown the analogy with optical instru- is well known to everybody. Given" electron lens " were eouh for ments. Whereas in the older type of the initial data, i.e., the position andthem to ask themselves :-nought can electron optical instruments the main bearing of the gun and its " charge "we do with electron lenses ? Can we problem was to bring electron trajec- which determines the initial velocity,make electron microscopes, electron tories to a focus in space, new devices from these data the whole trajectorytelescopes, etc. ? Within a few years have now come into existence, in can be calculated, among other thingsof Busch's paper Knoll and Ruska which electrons are brought together also the final point at which it strikeshad made the first electron microscope in space and time. This new sort of the ground. The correspondingand .Zworykin created the firsttube. focusing has been termed " phase Hamiltonian problem, however, is astical high vacuum cathode-ray tube. focusing " by Briiche and Recknagel. follows :-We consider as given theFifteen years of vigorous develop- I prefer " space-time focusing," as I initial point (the position of the gun),went followed. Looking back on these think this makes it at once clear what the final point (the target), and thefifteen years, I can scarcely remember

t Strictly speaking, Hamiltonian ideas, not total energy of the projectile (which isa paper dealing with any - practical methods, as for the purpose of this lecture it will beaspect of electron optics in which sufficient to explain the refractive index in more : These are strictly speaking always " many

detail than usual, but it will not be necessary to electron problems " and could not 'be treated by the* B.T.-H. Research Laboratory. introduce Hamiltonian devices such as the"Eikonal," usual " one electron " methods.



given by the charge). The question isnow :-What will be the trajectory ortrajectories of the projectile betweenthe gun and the target for a givencharge ? The solution will give amongother things the elevation of the gun,which in the Newtonian treatmentwas considered as part of the initialdata.

The answer is given by the Prin-ciple of Least Action, which is aspecial case of Hamilton's Principle,Among all the curves which connectthe gun and the target the projectilewill choose the one along which the" action " is minimum. The "action"is the integral of the kinetic energyover time

me.dt

and as the velocity v=dsklt, this canbe immediately transformed into aninter ral over the path ds.

v.ds.

The velocity of a projectile in thegravitational field depends only onthe initial velocity (i.e., on thecharge), and on the height above theinitial point. As we have consideredthe charge as given, we can say thatv is a function of the position of theprojectile only. Therefore the" action " integral can be calculatedat once for all imaginable paths con-necting the gun and the target.

Polar Diagram ofAction

fitelsfor Target MadeMaximum Range

Fig. I. The balllitic problem as illustration ofHamilton's Principle. Among all imaginabletrajectories only those can be realised forwhich the " action " is a minimum. The twodynamically possible trajectories are shown in

continuous lines.

To illustrate this, in Fig. r thefamily of parabolas is drawn whichconnect the two points. Of course wedo not know a priori that the trajec-tory will be a parabola, but I have as-sumed this to simplify the illustration.We can now calculate the " action "for every one of these parabolas, andrepresenting this in a polar diagram

as a function of the elevation of thegun we, obtain a curve with twomaxima and two minima. If the pro-jectile were strung on a wire connect-ing gun and target, it could travelalong any of these parabolas. (Up toone which reaches a -maximum height,beyond which the projectile wouldglide back into the gun. The diagramstops at this point; beyond it theaction would become complex). Butas the projectile is free, not strung ona wire, it can travel only along thetwo (continuously drawn) parabolasfor which the action is minimum. Asis well known, these two dynamicallypossible trajectories start with eleva-tions symmetrical to 45°.

Sir William Rowan Hamilton gavea beautiful optical illustration to thisrather dry mathematical theory. (Curi-ously, though the dry mathematicalskeleton was remembered, the opticalillustration was forgotten for almosta hundred years, until Schrodinger re-vived it). Hamilton noticed theanalogy of the Principle of LeastAction with Fermat's principle, ac-cording to which the light ray be-tween two points travels in"the short-est or the longest time. Let u be thevelocity of light in a medium with re-fractive index n, and c the velocityin vacuo, n = c lu. Fermat's prin-ciple states that

dt dslu = n.ds

must be a maximum or a minimum.Comparing this with the Principle ofLeast Action we can say at once, -thatthe projectile will travel on a trajec-tory which could be chosen by a lightray in a suitably graded atmosphere,with an optical density gradually de-creasing with height. The gradationmust be such that at every point therefractive index n is proportional tothe vehicity of the projectile, v. (Ata given " charge.")

One important difference must benoticed however. We see from thepolar diagram in Fig. i that if werealised such a stratified atmosphere,an observer would see four mirages ofthe gun, corresponding to the twominima and the two maxima. (Moreexactly : The observer would see thegun once horizontally, in its realposition, and in addition he would seethree mirages). But only two of thesepaths, the ones corresponding tominima can be realised in the caseof the mechanical problem.

We can now see at once why theHamiltonian viewpoint is so eminentlysuited for the kind of problem likelyto be encountered in electron optics.First of all the optical analogy pro-vides us with a complete terminology.

If it had done nothing else, this alonewould be an outstanding achievement.Think of it what it means if we can doour thinking in terms of " lenses "instead of for instance " axially sym-metrical bilinear transformations " !Mathematicians may be able to dotheir thinking in such terms, ordinaryphysicists and engineers certainly cannot. But apart from this, theHamiltonian method provides also themost suitable mathematical system toformulate many problems (particu-larly focusing problems) even if thesolutions are often reached in a morepedestrian way.

We can illustrate this immediatelyon the ballistic example if we ask thequestion :-" Can the gun focus itstrajectories ?" Yes it can, if the tar-get is at the maximum for a givencharge. (Fig. 2). If the elevation ofthe gun is approximately 45°, smalldeviations from this angle will 'causeonly negligible deviations at the tar-get. This is a well known result, onlywe are not- used to thinking of it as" focusing effect of the gravitationalfield." In the polar diagram of theaction a very flat minimum appears,in which one maximum and twominima merge' into one. We cat im-mediately generalise this :-Focusingwill occur if the action has a minimumof a higher order.

The Electron -Optical Refractive IndexThe methods applied in the bal-

listic example can be transferred with-out any essential change to the caseof an electron (or any charged par-ticle), in an electrostatic field. The'reason is that both the gravitationaland the electrostatic field are (scalar)potential fields. Consequently the

Polar Diagram of Action

for Targetat Maximum Range

Fig. 2. Focusing effect of the gravitationalfield. If the gun elevation is 45° the range ismaximum, and small departures from theelevation will not affect it. The focusing effectreveals itself in the action diagram as a very

flat minimum.

" refractive index " of an electrostatic.field will be proportional to the veloc-ity v of the electron, and independent



-Direction of Vector Potential --

Fig. 3. Electron -optical refractive indite .Top figure :-Electric field. The refractiveindex is independent of direction, the n -surfaceis a sphere. Bottom figure :-Magnetic field.The n -surface is still a surface of revolution,but no more a sphere. It becomes increasinglyasymmetrical with increasing magnetic field.In strong fields the index becomes negative in

a certain angular range.

of its direction. It can be representedin a polar diagram as a sphere.(Fig. 3).

We have always qualified the de -f nition of the refractive index by theclause " at a given charge." But thisrestriction is so natural in most elec-trostatic problems, that it can bealmost forgotten. In practically allelectron optical devices the electronsource is a low temperature hotcathode, which emits electrons withvelocities of the order of a tenth of avolt. As we have usually acceleratingpotentials of hundreds or thousands ofvolts, we can often neglect the smallinitial velocity and consider thecathode as emitting the electrons withzero ballistic charge. We must not,however, forget the initial velocityaltogether, it has some very importantconsequences.

The electrostatic field behaves there-fore " optically " like an isotropicmedium with varying density, e.g.,like a gas or liquid. It is very dif-ferent with the magnetic field. I cannot dwell here on the derivation of therefractive index' in this case. (It isbased not on the Principle of LeastAction, but on the more generalHamiltonian Principle, with theLagrangian function instead of thekinetic energy). I can state onlywithout proof that it is

e

n = v - - A.cos (A.v.)nt.t

where A is the absolute value of thevector potential of the magnetic fieldand (A,v) is the angle between thevectors A and V. Representing thisagain as a surface, with radii pro-portional to n, we obtain a very queershape. The refractive index in mag-netic fields is often dismissed in,

treatises on electron optics as "depen-dent on direction, as in crystaloptics," but this is rather covering upthe difficulties of the question. Incrystal optics the n -surface is neverworse than an ellipsoid ! The mostsingular feature of the magnetic re-fractive index is, that it is not thesame in opposite directions. This hasno counterpart in ordinary optics.Light can always travel back on thesame path on which it came, but notan electron in a magnetic field ! Itis well known that the magnetic forceon an electron is proportional to thevelocity vector and at right angles toit. If therefore an electron travellingone way in a magnetic field is de-flected clockwise, it will be'deflectedcounterclockwise if it is coming theopposite way. The assymmetry ofthe refractive index expresses thisfundamental (mis-)behaviour of themagnetic force. Moreover, the refrac-tive index can become negativeLooking at Fig. 3 it can not surpriseus that hardly anybody has evertreated a magnetic problem withoptical methods.*

Electrostatic Lenses-The Hamil-tonian Analogy immediately explainsthe existence and properties of elec-trostatic lenses, of which an exampleis shown in Fig. 4. As we have seen,the refractive index is proportional tothe electron velocity, and this againis proportional to the square root ofthe potentialt measured from thecathode as zero level. All we have todo is to imagine the field replaced bya medium of continuously increasingdensity. This is shown inincreasing density of the shading.We can therefore consider the field asa succession of infinitesimally thin

* I say " optical " not " Hamiltonian." Thegeneral Hamiltonian theory of magnetic fields hasbeen developed by W. Glaser.

t This is true only for moderate voltages. Atnear -light speeds the refractive index increases fasterthan the square root of the voltage.

Trqiectory viewed indirection of axis

Angular Momentum p.v.. Magnetic Fluxthrough the surface of revolutiongenerated by the trajectory.

Fig. 5. Magnetic electron lens. Proof that atrajectory starting from a point of the axis wi I

again intersect the axis.

Fig. 4. Electrostatic electron lens. Thepotential increases from left to right. Anexact optical replica could be constructed byfilling the space between equipotential surfaceswith materials of optical density proportionalto the square root of the voltage. Acondensing and a dispersing lens can be usedas rough illustration, as shown in the bottomfigure. The condensing lens is always stronger

than the dispersing lens.

lenses, each bounded by a couple ofequipotential surfaces. As a combina-tion of lenses is again a lens, we havea full qualitative explanation of theelectrostatic lens,

Magnetic Lenses-The lens effectis far less evident in the case of a" magnetic lens," such as can be pro-duced by the field of any axial coil.(Fig. 5). The magnetic field exerts aforce on the electron at right anglesto the field lines and to the movementof the electron. It is evident that thetrajectory will circle round the axis ina sort of spiral. The question im-mediately arises If an electronstarts from a point of the axis, willthe trajectory return somewhere to theaxis, or will it miss it ?"

It would not be profitable to applysimple optical considerations to thiscircling movement of the electron asin every plane at right angles to theaxis the refractive index has the queershape illustrated in Fig. 3. For oncewe borrow, without proof, a resultfrom dynamics, which fully describesthe movement of the electron at rightangles to the axis. This is asfollows :-

Let us rotate the trajectory roundthe axis. We obtain a surface of re-volution. Between two points on itstrajectory an electron acquires anangular momentum round the axiswhich is proportional to the totalmagnetic flux through the part of thesurface, of revolution which is definedby the two points.*

If an electron starts anywhere fatoutside the field and leaves it again,the resulting flux through the en-

* This theorem was derived by Busch, but firstexpressed in these words by Bouwers.



Lines A =Const.not field lines !

Refractive Index In Meridian Planerotating with electron

a A2n= 4,k7-

2m If V

Fig. 6. Magnetic electron lens. Abstractingfrom the rotation of the trajectory around theaxis the lens effect can be expressed by anisotropic refractive index, as in the case of theelectrostatic lens, but with a different formula.The optical replica would have a densitydistribution as indicated by varying degrees of

shading.

velope of its trajectory is zero, itstotal gain of angular momentum istherefore nil. If we consider an elec-tron which has started from the axis,i.e., with an angular momentum zero,it will have again zero momentumwhen leaving the field. This meansthat after leaving the field the move-ment of the electron will be strictlyin a meridian plane, and either thepath itself, or the tangent to it willintersect the axis.

This theorem, which we had toborrow from dynamics enables usnow to return to the optical analogy.We consider the movement of theelectron in a meridian plane whichturns with the trajectory round theaxis. In this rotating meridian planeit- is quite profitable to apply opticalconsiderations. We have seen in Fig.3 that though the surface representingthe refractive index in a magneticfield has a very strange shape in anyplane containing the vector potential,every section at right angles to thevector potential is a circle. But thevector potential of an axial coil runsin circles round the axis, therefore wecan interpret the movement of anelectron in the (rotating) meridianplane as caused by a refractive indexof the familiar isotropic type. (Fig.6). Taking the rotation of the tra-jectory into account one obtains forthis an expression.

e' A'v - --m2 v

or, expressing the electron velocity vby the electrostatic potential V, andleaving away an irrelevant constantfactor

e A'n.,

2m V V

In the purely magnetic lens V isconstant, therefore the lines of con-stant refractive index will be the linesA = const. These lines are similar tothe magnetic field lines, but not iden-tical with them. On the axis A isalways zero, therefore the magneticrefractive index will always decreasewith increasing distance from theaxis. In Fig. 6 this is again indicatedby shading with varying density. Thisresult may be of some practical irriportance, as it allows the transforma-tion of a magnetic field into anequivalent electric field andmakes it possible to trace trajectoriesby well known methods. It must benoted, however, that the formula fornm is valid only for electrons whichhave crossed the axis, or which havest,rted from a cathode outside themagnetic field.

This figure shows clearly, and prob-ably somewhat surprisingly,. that theoptical analogy of a magnetic lens isvery unlike anything which we areused to call a " lens " in optical prac-tice. There is nothing of the lentilshape about it, from which the word" lens " is derived. On the contrary,the field is built up of thin hyperbo-loidal shells, with refractive indicesgradually decreasing. Near the axisthe decrease is approximately propor-tional to the square of the axial dis-tance. Though this is a very unusualtype of lens, its'effect is the same as ofthe familiar type. The electron willrun along the axis on a potentialcrest, which acts on a negative chargelike a trough for a heavy body. If itdeparts from the axis, the slopes atboth sides of the trough will drive itback towards it. We understand atonce why the magnetic lens actsalways as a converging, never as adiverging lens. (The correspondingtheorem for the electrostatic lens isnot so easily proved).

It may be added, that the opticalanalogon could be made perfect if wemade up a model out of transparenthyperboloids and rotated them at ex-treme speed. In such a rotating lenslight would describe just the samesort of spiral path as an electron in amagnetic lens. There is, however,little hope of realising such a modelexperimentally as the convection oflight by moving media (Fizeau effect,really a relativistic effect) becomesnoticeable only at speeds at which anymaterial would burst by centrifugalforce.

Lens Errors.-As soon as Busch hadproved that axially symmetricalfields, electric or magnetic, behave inthe first approximation as lenses, aquestion very naturally suggested it-self : " Can we make electron lensesas perfect as optical lenses ?"

The question could not be answeredoff hand; many years of work wereneeded before it was fully elucidated.As compared with light optics elec-tron optics started with a great ad-vantage, and with a great disadvan-tage. The great advantage was thatin electron optics we can realise re-fractive indices far larger than inlight optics. In optical instrumentthe ratio of the largest and smallestrefractive index is never more than2 in electron optical devices wecan, easily realise ratios of r,000

On the other hand electron opticsstarted with a great disadvantage.Modern applied optics has a greatvariety of optical glasses (and nowa-days also plastics), at its disposal,with a considerable variety ofchromatic dispersion coefficients.Electron optics has only two"media," the electric and the mag-netic field, both with fixed dispersionproperties. Nor can we shape theelectron optical lenses as freely as theoptician shapes his lenses. Weare limited by the laws of zerodivergence for the electric field, zerocurl for the magnetic field. (Freedomfrom space charges and spacecurrents).

It will be useful for the understand-ing of the following explanations 'ifwe anticipate the results, and sum upthe essence of 15 years of develop-ment in a few words :-

It was not possible for the scienceof electron optics to design opticallyperfect or at least highly correctedlenses, not because of lack of abilityof the theoreticians, but because ofthe laws of the electro-magnetic field,The disadvantage which we men-tioned above was too great. On theother hand electron optical practicecould overcome all .practical difficul-ties by availing itself of the great in-

SPHER/CAL ABERRATION

srfaster sketrori

afowsr Istm.CHROMATIC' ABERRATION

Fig. 7. Lens errors for axial object points.

* As the initial energy of the electrons emitted bya barium cathode is of the order o.x volts, this canbe realised with about roo,000 volts acceleratingvoltage.



herent advantage of electron optics,by the use of high voltages. The besttheory could do was to point out theessential limitations, and to save thepractical workers in the field fromwasting time and energy in trying. toreach unattainable ideals. In the shortsurvey which follows, I want there-fore to lay particular emphasis onthese theoretical limitations.

The first question which arises is :" Can an electron lens system imagea point of the axis in a point ?" Inthis case there are two defects pos-sible, spherical aberration andchromatic aberration. Both are illbs-trated in Fig. 7. Scherzer has shown inan important paper' that neither canbe eliminated Trajectories starting ata larger off -axis angle will alwayscross the axis before trajectoriescloser to the axis. On the other handfaster electrons starting at a givenangle will always cross the axisbeyond slower electrons with the sameinitial tangent. The deeper reasonfor both defects is that it is impossibleto realise electromagnetic diverginglenses.

Electron MirrorsA certain possibility of overcoming

these defects emerged in 1936 whenHottenroth and Recknagel' (AEGLaboratory) discovered that electronmirrors may have aberrations op-posite to those of lenses. Such mirrorscan be realised as shown in Fig. 8,but also with electrodes of the sameshape as in lens arrangements, simplyby giving the second electrode apotential sufficiently negative to repelelectrons, and to return them towardsthe same side from which they came.Fig. 9 illustrates that spherical andchromatic aberration in the case ofmirrors, and shows that they mayhave signs opposite to those of lenses.By combining lenses and mirrors itmay be therefore possible to produceaberration free systems.

But this was a forlorn hope, forseveral reasons. In the case of mir-ors, the image appears at the same

side as the electron beam arrives.There are two possibilities to over-come this. One is to use two mirrors,as in the Cassegrain systems of tele-scopes. But this has two difficulties :First, the second mirror creates avery bad disturbance in the arrivingbeam, by its edge field and by the,struts which are necessary to hold it.(There is no electron optical equiva-lent to " painting the struts black.")Second, this second mirror, or imagescreen, cuts out the paraxial rays, andforces the designer to use trajectoriesso far from the axis, that it is not suf-ficient to correct the system to thethird order. Even in light optics suchmirror systems can be profitably

Fig. 8. Electron mirror. Similar effects canbe obtained also with electrode gementsas in Fig. 4, if the second electrode is strongly

negative against the cathode.

realised only with non -spherical sur-faces. There is . also another pos-sibility, viz., to incline the mirror.But skew (non -centred) systems aregenerally so difficult that even in lightoptics advances in this direction havebeen made only recently.'

Apart from these theoretical diffi-culties, electron mirrors offend badlyagainst the fundamental practicalprinciple of electron optics (of whichmore later) :-" Do not slow downyour electrons if you can help it !"No wonder that images obtained withelectron mirrors are mostly of ex-tremely bad quality, and the hope of

Spherical Aberration of Mirror

s/ortsr electronfaster electron

Chromatic Aberration of Mirror

Fig. 9. Axial aberrations of electron mirrors.

obtaining fully corrected systems withelectron mirrors must be classed as anillusion.

There is also another possibility toeliminate the spherical aberration, byspace charges. There seems to be afairly widespread belief, that thespace charge of an electron beam willalways prevent its focusing in apoint. This is true, of course, for abeam which would be stigmatic with-out a space charge. But generally thespace charge produces an error of

opposite sign to the spherical aberra-tipn, and it is possible that under cer-tain conditions the two just cancel oneanother* It can be easily shown thatthis is not impossible, as the energystored in a perfectly stigmatic beamcarrying a finite current is finite. (Per-haps it may be worth while to refutethe fallacious belief that if an elec-tron beam passes through a point, theelectrons much touch each other.They will pass in single file !) Butthis remedy has no practical import-ance. It can not be applied in elec-tron microscopes, where the beam hasan incalculable shape and density dis-tribution, depending on the object,and in cathode-ray tubes, for reasonsto be discussed later the sphericalaberration presents no seriousproblem.

We see that the few remedies so fardiscussed are useless. Yet, practicehas overcome the lens defects, even in

.the case of the only device in whichthe spherical and the chromatic aber-ration threatened to be dangerouslimitations, in the electron micro-scope. I have mentioned already thesimple panacea of practice : highvoltages. In an electron microscopeobjects are made visible by electronabsorption and scattering. If thebeam voltage is raised, absorption be-comes negligible and the energylosses of the electrons become less andless, not only relatively, but even ab-solutely. The scattered beam becomesincreasingly homogeneous, and thechromatic effect becomes negligible.As regards the spherical aberrationand also other lens effects, they arealso reduced because the scatteredbeam is confined to a rather narrowangle. (Further practical remedieswill be discussed in connexion withthe electron microscope).

Theory could give only one helpfulhint, and this came from R. Rebsch,'a pupil of Prof. Scherzer. Rebschshowed, that although the sphericalaberration could not be eliminated, itcould be reduced in electron micro-scopes to any insignificaht value, byallowing the field of the objectivelens to extend to the object, ina way which is equivalent to pushinga very small, yery weak lens veryclose to it. This artifice is useful alsoin the case of electrostatic microscopesimaging the cathode, where highvoltages would not bring anyimprovement.

(To be continued).

1 0. Scherzer, Zeitschr. f. Physik lox, 593, 1936.2 G. Hottenroth Zeitschr. f. Phys, 103, 460, 1936.

A. Recknagel, ZS. f. techn. Physik, zy, 643, 1936.3 Mainly through the work of Dr. C. R. Burch

and Dr. E. H. Linfoot of Bristol.4 R. Rebsch, Ann. d. Phys, 5, 31, 551, 1938.

* For other possible beneficial effects of the spacecharge cf. 0. Klemperer, Brit. Pat. 534,215.



Frequency Mixing with a Triode-HexodeBy E. HUGHES, D.Sc., and E. F. PIPER, A.M.I.E.E.*

WHEN two sinusoidal voltages offrequency 11 and h are addedtogether, the amplitude of the

resultant wave varies between a maximumand a minimum at the beat frequency of(f, -12) ; but a beat -frequency componentis not obtained unless this resultant waveis either partially or completely rectified.It is of interest to examine how suchpartial rectification is obtained with thetriode-hexode, and to deduce the factorsthat govern the amplitude of thebeat -frequency component.

Graphs showing the variation of anodecurrent with control -grid voltage for atriode-hexode are reproduced in Fig. 1.

- Each graph corresponds to a givenoscillator -grid voltage. It will be observed(a) that the graphs are linear over aconsiderable portion of their lengths, (b)that for a given control -grid voltage, thevertical distances between the 'graphs arepractically proportional to the variationof oscillator -grid voltage.

Suppose the control grid to have anegative bias, Vb, and the oscillator grida negative bias of, say, 2 volts.Consequently, the valve is being operatedabout point Q in Fig. r, and it is evidentthat an alternating voltage of limitedamplitude applied to either the controlor the oscillator grid causes a correspondingundistorted variation of the anode current.But when sinusoidal voltages of slightlydifferent frequencies are applied to thetwo grids simultaneously, the anodecurrent is distorted as shown in Fig. 2.This waveform is the oscillogram of p.d.across a resistance R (Fig. 3) of r,000 ohmsinserted in the anode circuit whenvoltages of about roo and 124 kc/s areapplied to the control and oscillator gridsrespectively. It will be seen that themean value of the anode current variesat the beat frequency of about 24 kc/s.

Let us now consider how this waveformcan be derived from the characteristicsshown in Fig. 1. It is found that thestraight portions of the graphs, whenproduced, converge to a point P on ornear the X axis.

Suppose the valve to be operated aboutpoint Q and an alternating e.m.f. ofamplitude Vo and frequency h to beapplied to the _control grid only ; then,using the symbols shown on Fig. 1,

instantaneous anode current =ii

=l - Vb + Vo sin zinc t) - d I (I)where g = slope of graph PQ = a/c.

Next, suppose an alternating voltageof amplitude Vo and frequency foto be applied to the oscillator gridsimultaneously with the above -mentionedvoltage to the control grid. Then theslope will vary between a maximumcorresponding to graph PM and aminimum corresponding to PN ;

a + bi.e., maximum slope = , and

C

* Technical College, Brighton.

Q11-) 7BL

i0000n Icn-2500

Fig. I. (right) Characteristics of thetriode-hexode.

Fig. 3. (above) Circuit used for ex-perimental verification of ex-

pression 2.

minimum slope =a - b

J_

LDl0000

2500C

cSince the intercept is practically

proportional to the value of V0, b, = kVo,where k is a constant for a given valveoperating under given conditions.

Hence, instantaneous slopea ± k Vo sin 2 in fo

C.and the general expression for theinstantaneous value of the anode currentis obtained by substituting for "g " in(I), thus :

- kV, sin 27rfo ti

C

a k (c - Vb) Vo a= {- (c--- Vb) - d} sin 27rfo t + Ve sin 27rfe t

- V, Ve sin 27rfe t. sin 21r fe

(3)

-c

CONTROL GRID VOLTAGE

ANODECURRENT

T 6

two applied voltages and independentof the biases applied to the controland oscillator grids, so long as thevalve is being operated within thelinear limits.The conversion conductance, namely,the change in anode current at beat(or intermediate) frequency per volt

kV,applied to the control grid, is

2Cb -

i.e., - in Fig. 1. Since the maximum,2C

[C - Vb + Vc sin 2irfe t]) - d

k (c - Vb) Vo= q+ sin 2irf. t,+ g V, sin 21rle

+ - Vo Vc {COS 27 (fo - fc) t - COS 2TT (Jo + fe) t2C

where q = g (c - Vb) -d= anode current with no

alternating voltages applied tothe control and oscillatorgrids.

From expression (2), it follows that :

(I) The anode current has componentsof frequencies fo, fo, (fo - fo) and(.fo fe).

(2) The amplitude of the componenthaving frequency (fo - fo) in.proportional to the product of the

. (2)

1

Fig. 2. Waveform of anode current with volt-ages of different frequency on control and

oscillator grids.



I - rfLCIni

VIMGO.TRAD.MARK) 'FAHIGH FREQUENCYINSULATION

MYCALEX is unaffected by tempera-ture up to 400° Centigrade. It is

tough and unaffected by water, gas oroil, yet can be machined to fine limits.In addition it can be combined withmetal as in the example shown below(right). Components of this type arein production for purposes where heatconditions have to be met.

MYCALEX sheet cut, milled, tre-panned, tapped, without fracture orcrumbling, and all edges sharp and true.

A copper inserthas become anintegral part ofthe whole throughclose adherence tothe MYCALEXsurround.

Our experience is at your service onapplication to

MYCALEX LTD.CIRENCESTER, GLOS.

LM

TELCONCABLES

for

HIGHFREQUENCY

TRANSMISSIONand RECEPTION

Full particulars of all Telcon Products from :

THE TELEGRAPH CONSTRUCTION& MAINTENANCE CO. LTD.Head Office : 22, Old Broad Street, London, E.C.2

Telephone : London Wall 3141

Askilled team of workers in an efficiently equippedmodern factory produces to an unusually high standardall types of metal instrument cases, radio chassis,panels, brackets, boxes and othet metal components.We are also able to undertake the design and manu-facture of complete amplifying equipment for special

requirements.

PRIORITY ORDERS ONLY

ALFRED Rit Of ETD. 112-06,New Oxford Street,london,WC.



Grid Voltage

I

(b.H

value of b cannot exceed a, it followsthat the conversion conductancecannot exceed half the mutualconductance, which is a/c.

Conclusions (r) and (2) can be confirmedexperimentally with the aid of the circuitshown in Fig. 3. Oc and 00 are twoAvo-oscillators ; L1 and L2 are suitableinductances, very loosely coupled ; andR is a relatively low resistance of, say,1,000 ohms, short-circuited by switch Sexcept when it is used to give thewaveform of the anode current. Non -inductive resistances of 2,500 and ro,000ohms are connected in series across eachoscillator output to enable the voltageapplied to the respective grids to be eachincreased five -fold.

Suppose the frequency of O, to be r Lc)and that of 00 to be 250 kc/s. We shouldbe able to detect in the anode current thepresence of these two frequencies, togetherwith frequencies of 149 and 36o kc/s.The simplest procedure is to adjust 00(with 00 off) to each of these frequenciesin turn ; and for each frequency, to notethe scale readings on C1 and C2 for themaximum output on the cathode-rayoscillograph, CRO, connected across L2C2.

The oscillators 00 and 00 are then setfor rro and 25o kc/s respectively, and itis found that each of the four frequenciescan be tuned in on L1C1 and L2C2, therebyconfirming the first conclusion.

The multiplicative action was checkedby tuning L1C1 and L2C2 to the beatfrequency of 140 kc/s, varying theamplitude of the voltages applied to thecontrol and oscillator grids and notingthe perpendicular distance between thepositive and negative peaks on theoscillograph screen. The results obtainedwith an X24 triode-hexode -are given inthe table in the next column :-

Time

(d.)

Fig. 5. Graphical construction ofanode current waveform with Vgin phase (c) and in anti -phase (d).

ProductControl- Oscilla- of Vertical

grid tor -grid relative distancetapping tapping ampli-

tudesbetween

peaks

C (r) E (r) r 0.15 inchB (5) E (r) 5 0.7 inchC .(r) D (5) 5 0.8 inchB (5) D (5) 25 3.5 inch

The figures in brackets represent thecorresponding relative amplitudes of theapplied voltages. It is seen that theoutput beat -frequency voltage ispractically proportional to the productof the input voltages, and the multiplica-tive action is therefore confirmed.

When performing this test, it isimportant to check that the resistancesused as potentiometers operate satis-factorily at the frequencies employed.This can easily be done by setting eachoscillator in turn to 140 kc/s (the otherbeing off), and noting the output voltageson the CRO when the contacts are on Band C or on D and E.

Fig. 4. Variation of out-put voltage with varia-tion of control grid bias(curve `A)tand oscillator

grid bias (curve B)

With reference to the second part of- conclusion (2), it is not obvious from thevalve characteristics that the amplitude

_ of the beat -frequency component shouldbe independent of the grid biases-especially the lias, on the control grid.With the potentiometers shown in Fig. 3,it was possible to vary these biasesindependently, the LC circuits beingtuned to the beat frequency of 140 kc/s.The variations of the output voltage areshown in graph A being for thecontrol grid and B for the oscillatorgrid. It should be pointed out thatfor the anode and screen voltages usedin this test, the graphs of Fig. r beganto depart from linearity when thecontrol -grid bias exceeded about -2volts,while the vertical spacing per voltbetween the graphs increased slightlyas the oscillator -grid bias was increasedfrom o to -4 volts. Hence, it followsthat within the limits of linearity, thewhole of conclusion (2) can be confirmedexperimentally.

It is realised that the abovemathematical proof neglects the effectof load impedance. The dynamicimpedance of L1 C1 (Fig. 3) 'is negligiblecompared with. the anode a.c. resistanceof .the mixer at all frequencies exceptwhen L1 C1 is in resonance. Evenfor this condition, the dynamic impedanceis usually relatively low and thereforedoes not greatly affect the anode current.This can easily be demonstrated byconnecting the CRO across R (with Sopen) and noting the waveform as C1is varied through resonance.

It is of interest to apply expression (2)to the case where the two frequenciesare equal and where the voltages appliedto the control and oscillator grids are(a) in phase, and (b) in anti -phase.

When the control and oscillator gridare in phase, it follows from (2)

that-b (c- Vb)

i = q gVe sin 2/rft

bVccos zir (V) t)

2C t

(3)

B

-4 -3 _ -2GRID BIAS VOLTS

3"

rn

2

O

OZ



Hence the effect of the I appliedalternating voltages is to increase themean value of the anode current by

b

2C

The effect of reversing, say, theoscillator -grid voltage relatively to thecontrol -grid voltage is to reverse thesign in front of " b " in expression (3).Hence, with the voltages in anti -phase,

b(c - Vb)= q tgVe I sin 27rft

c )

bV,- - cos 2ir (2f)2C

(4)

Hence the effect of applying the voltagesin anti -phase is to reduce the -meananode current by bile/2c comparedwith the initial value q. The conversionconductance, 13/2c, is therefore givenby the change of the mean anode currentwhen a p.d. of 1 volt peak value is appliedto the control grid, or by half the changeof mean anode current when that p.d.is reversed relatively to the oscillator -gridvoltage (see Rapson's Experimental

Radio Engineering, p. 34).Expressions (3) and. (4) indicate the

introduction of second harmonics intothe waveforin of anode current. Thiscan be confirmed experimentally bynoting the waveform of p.d. across alow resistance in the anode circuit orby the graphical construction Shown inFig. 5. The valve is assumed to bebiased so as to operate about point Q ;and the oscillator grid voltage is assumedto be such that when it is at its maximumpositive value, the anode current lieson graph CM, and when at its maximumnegative value, it lies on BN. Bydrawing intermediate graphs such thatQF = QE sin 3o° = 0.5 QE, andQG = .QE sin 6o° = o.866 QE, etc.,it is possible, by projecting upwardsfrom the curve of control -grid voltageand then horizontally, to deduce curve (c)of anode current when the control andoscillator grid voltages are in phaseand curve (d) when they are in anti -phase.Oscillograms of p.d. across a low resistanceinserted in the anode circuit of atriode-hexode, the oscillator and controlgrids of which were connected to separatesecondaries of a low -frequencytransformer, were in close agreementwith the curves deduced graphicallyas in Fig. 5.

It was found that when two separateoscillators were used, it was much moredifficult to get oscillograms of anodecurrent showing waveform (d) of Fig. 5.This particular shape was very sensitiveto change of oscillator frequency andto adjustment of the time and synchronis-ing controls on the CRO.

All experimental evidence confirmedthat expression (2) does represent correctlythe factors that govern the operationof a triode-hexode when used as afrequency -mixer.

MODERN DESIGNBy WHITNEY S. GARDNER

You have removed the chassis fromyour receiver to replace a resistor.That job done, it simply becomesnecessary to slip the chassis inside thecabinet and attach a few knobs. Acinch.

You place the chassis squarely onthe table and, holding the cabinetfirmly between the left and right hands,slowly lower it. The front bottomedge strikes the condenser shaft, so youease the cabinet forward. Again youlower.: it. :This time the rear edge ofthe cabinet strikes the power transformer.

Remove the cabinet. Study thesituation. Remove the tubes to get moreclearance. Lower cabinet. Ah, it clearsthe transformer, but strikes the condensershaft. Remove cabinet and place itupside down on table. Invert chassisand lower it into cabinet. There, justtip it slightly and it clears both transformerand condenser shaft. Oh, oh-it slipped.Now it's gone down too far. Invert thewhole works. Now lift up on cabinet.You can't. That's because condensershaft and transformer have -a tight gripon cabinet. Front of cabinet bulges.Bang it sharply on table. Cabinet givessuddenly and comes off chassis. So doesdial pointer.

Straighten dial pointer. That's goodenough.- Lower chassis into cabinet,but not too far. Something is holdingit back. Why, of course ; the powercord should have been threaded throughthat grommet first. Now once more.At last it's down there, but the condensershaft is still causing the cabinet to bulge.That's because it is not in line with thehole. It should be moved one-half inchto the left. It doesn't slide. Better takeit off again. That time you only rippedoff- the dial light wire.

Resolder it. Take a rest. That tincabinet was on there once and it isn'tgoing to whip you. There, on it goesand all shafts lined up. Just goes toshow what clear, calm thinking will do.Attach knobs. Attach aerial and ground,plug in power cord and snap on the power.Nothing happens. That's because youforgot to put the dial light back in thesocket. Oh, well, you always have theroom light on anyway. But wait, nohum, no hiss. Well, naturally-you tookout all the tubes, remember ?

That narrow opening in the back isjust large enough for your hand. There-six of them are in. You have one left.That's the one that goes up on the shelfnear the dial light. You can't reach it.Someone put it in there ; you ought tobe able to. The awful truth dawns. Ithas to be inserted before the cabinet isattached !

" Hey, Marjie, what picture are theyshowing at the Strand to -night ? "

-Q.S.T. Vol. 25. No. 9. September,1942, page 18.

SELENIUMPHOTOCELLS

The " EEL " photo cell has noequal- in sensitivity and stabil-ity. They are made in standardsizes of 25, 35, 45 and 67 mm.diameter and of 4 x 8, 11 x 18,37 x 50, 22 x 40, and 17 x 43.5m/ m. in the rectangular. Allcells are guaranteed (accordingto size) from 0.50 to 0.65 mic-roamps per ft. candle per sq. c/rn.

For full details write to:

EVANS ELECTROSELENIUM Ltd.

Bishops Stortford

Herts,

THE SEAL OFQUALITY



Taking Care of TransmittersThe following suggestions for the care and maintenance of transmitting valves are takenfrom a useful booklet: " Prolonging Tube Life," issued by the Heintz Kaufman Co., California.The original is illustrated with amusing sketches of which three are reproduced here.

THE life of a tube in normal ser-vice depends upon the numberof watts it is required to dis-

sipate on the plate. If the plate lossin watts is reduced, the life goes upproportionately. In other words, tubelife may be expressed as " watt-hoursof plate dissipation," and any reduc-tion in watts results in a gain inhours. Therefore, it is advisable toadjust every circuit so that the high-est efficiency is obtained.

Keep circuits properly tuned. A

small amount of detuning in theplate circuit causes a rapid increasein the plate dissipation of the tube.Circuits often detune as the transmit-ter heats up, and readjustment is thennecessary.

In Class B audio amplifiers, the" no -signal " plate current can oftenbe reduced without resulting in harm-ful distortion. This reduction savesprecious " watt-hours."

Minimise stray circuit losses inClass C r.f. stages, and make sure thatthe loading on the tube is useful load-ing. To test for this, disconnect the_ useful load and check the unloadedplate current.

The grid current of a triode is agood indicator of the amount of r.f.grid driving voltage required. Inordinary Class C r.f. amplifiers thegrid current should be roughly I to1/6 the d.c. plate current of the tube.For doubler or tripler service, wherelarge grid -leaks on the order of 50,000ohms are employed, the_ratio of gridto plate current may fall off nearer1/ to.

A good experimental way to adjustto the proper amount of grid drive,is to reduce the drive until the effi-ciency of the tube starts to fall off.This will be indicated by a visible

increase in plate heating. The griddrive should then be restored some-what above this fall -off point.

When the tube is idle the filamentshould be turned off. When both theplate and filament voltage can beturned on simultaneously, the fila-ment may be turned off in stand-byservice also, since a thoriated tung-sten filament is ready to operate inless than one second after the voltageis applied.

In these days no one can afford theluxury of an experimental set-up ora slightly " hay -wire " condition inthe circuits and power supplies of avacuum tube transmitter. Accidentalcircuit failures and accidental failuresof component parts, will often destroythe tube.

Avoid excessive grid drive. Excessgrid drive (grid current) wastes driv-ing power, and shortens the life of thedriver tube by making it do extrawork. Excess grid current also over-heats the grid of - the tube, andshortens its life either by damagingthe grid permanently, or by increasingthe number of watts the tube mustdissipate.

It is essential that the rated fila-ment voltages be maintained at thetube. This voltage should bemeasured at the socket, and shouldnot deviate more than plus or. minus5 per cent. from the rated value. -

The life of a thoriated tungsten fila-ment will be reduced to 2/3 ofnormal if the filament voltage is per-mitted to run to per cent. above itsrated value. At to per cent. belowrated value, the emission from thefilament may fall off due to failure todiffuse enough thorium to the surfaceof the filament to maintain emission.A drop in emission may cause severeoverheating of the plate, with a con-

sequent reduction in the life of thetube, or even complete failure.

Make good electrical connexions tothe tube. At ordinary frequencies, thestandard connector clips are satis-factory. At u.h.f. the charging cur-rents into the inter -electrode capaci-ties become large enough so thatspecial care must be taken. A splitconnector of aluminium or platedbrass, with the two halves held to-gether by a silver or similarly platedexternal spring, which will remaingood at 200 to 30o degrees C., willprove most satisfactory.

The electrical instability of r.f. cir-cuits increases the probability of dam-age and over -load to a tube. Parasiticoscillations can also cause damagingoverloads, as well as inconvenience.Nearly any parasitic oscillation canbe prevented. A good way to isolateand cure parasitics in an amplifier isto :

(a) Remove the normal excitation.(b) Remove all fixed bias.(c) Lower the plate voltage until

the plate loss due to the static platecurrent flowing does not exceed therated tube dissipation.

Under these conditions there shouldbe no parasitic oscillation at any posi-tion of the tuning dials. A parasiticoscillation will be readily indicated bythe presence of grid current. If suchoscillations occur, then

(a) Find the frequency of theparasitic.

(b) Determine the parasitic circuitsuperimposed on the normal r.f.circuits.

(c) Adjust the parasitic circuit de-creasing its excitation voltage untilthe oscillation ceases. Such changesneed not seriously affect normalcircuits.



Value provenby long service

nsrainfor

INDUSTRIAL APPLICATIONS

ELECTRO MEDICALEQUIPMENT

THERMIONIC INSTRUMENTSand TEST GEAR

VITAL COMMUNICATIONSand WARNING SYSTEMS

ETC.

THE GENERAL ELECTRIC CO: LTD.Magnet House, Kingsway, London, W.C.2



ABSTRACTS OFELECTRONIC LITERATURE

RADIOA Study of Propagation over the Ultra -Short Wave Radio Link BetweenGuernsey and England on Wavelengths

of 5 and 8 metres.(R. L. Smith -Rose, D.Sc., Ph.D., and

(Miss) A. C. Stickland, M.Sc.)Analysis of the data obtained has

shown that while the fading of thesignals was similar in type on the twowavelengths, there was a difference inthe secular variation of the amount offading on 5 and 8 m. No diurnal orseasonal variation of signal intensitywas noted, nor was there any de-finite diurnal variation in the amountof fading. The fading observed tookvarious forms, ranging from a rapidtype in which the changes occurredmore or less periodically every threeor four minutes, to a slow type ofvariation extending over one or morehours. The latter type was usuallyaccompanied by a high level of signalintensity interrupted at long intervalsby rapid and deep fading lasting fromio to 30 minutes. The field strengthrecords show a marked correlation be-tween periods of negligible fading andthe existence of low atmosphericpressure conditions generally, includ-ing lack of appreciable temperatureinversion. The occurrence of fog andsnow has a similar effect to that dueto the prevalence of low pressure.

-Jour. I.E.E. to be published.Transients in Frequency Modulation

(H. Salinger)In a frequency modulation system a

sudden jump in carrier frequency cor-responding to a Heaviside unit signalwill result in a transient depending onthe receiving -filter bandwidth. If thisbandwidth exceeds twice the maxi-mum frequency swing, the shape andduration of the transient is shown tobe about the same as in an amplitude -modulation system with the samebandwidth for narrower filters thetransient lasts longer.. These resultsare applied to several practical cases.The transient is favourably affectedby using an amplitude limiter and byarranging the filter pass band so as toenclose the maximum frequency swingsymmetrically.

I.R.E., Vol. 3o, No. 8page 378.The Technique of Frequency

Measurement and its Application toTelecommunications

(J. E. Thwaites and F. J. M. Laver)Rigid frequency control has become

a necessity in radio broadcasting andindeed in all forms of telecommunica-tion, and the art of frequency measure-

,

ment has likewise become of great im-portance. The improvement in fre-quency standards which has takenplace in recent years is briefly out-lined in this paper, and some of themethods which may be used to obtainaccurate frequency measurements interms of the improved standards aredescribed.

-Jour. I.E.E., Vol. 89, Part 3, No:7 (1942), page 139.

THERMIONIC DEVICESThe Use of Secondary Electron

Emission to obtain Trigger or RelayAction

(A. M. Skellett)The use of secondary electrons to

obtain trigger action similar to that ofa thyratron is described. An experi-mental tube and the necessary circuitsby which this action is achieved arediscussed, This combination givesthe features of a triode with a relay oron and off feature, resulting in anamplifier, oscillator, modulator orother vacuum tube device which maybe turned on .or off abruptly at high orlow frequencies. In addition, it canbe used to replace thyratrons in manyof their circuits where very low im-pedance is not necessary and is cap-able of much greater speeds of opera-tion in such applications.

-Jour. App. Physics., Vol. 13, No.8 (1942), page 519.

THEORYThe Characteristic Curve of the

Triode(E. L. Chaffee)

A method is described by which theentire static characteristic curves forthe anode current of a power triodecan be deduced from one experi-mentally determined curve. This ex-perimental curve can be obtained atlow power without danger of over-heating the valve. Using the sameprocedure, the grid current curves andthe total space current curves can bedetermined. A new log chart is des-cribed on which the static character-istic curves of a triode are presentedby means of straight lines and twocurves giving the division of spacecurrent between grid and anode.

-Proc. I.R.E., Vol. 30, No. 8

(1942), page 383.Simplified Inductance Chart

(E. S. Purington)A reference sheet in which the in-

ductance and Q of solenoidal or multi -layer coils wound on a cylinder arereadily determined when the dimen-sions or the coil are known. Effect ofinsulation in reducing space factor are

considered together with skin effectconsiderations.

-Electronics, Vol. 15, No. 9, p. 61.Variation of the Axial Aberrations of

Electron Lenses with Lens strength(E. G. Ramberg)

The variation of refractive powerand spherical aberration with elec-trode voltages and field strengths isstudied for two characteristic uni-potential lenses, an immersion lens,and a magnetic lens. Conclusions aredrawn herefrom regarding the varia-tion, with lens strength and appliedvoltage of the resolving power obtain-able with the lens as an electron -microscope objective. Scatteredmeasurements by other authors agreesatisfactorily with the results. The" relativistic aberration " of the elec-trostatic unipotential lenses, i.e., theeffect on the image of fluctuations inthe over-all applied voltage, is cal-culated and shown to be of significancein the electrostatic electron micro-scope. Furthermore, the axialchromatic aberrations are computedfor the four systems and the questionof upper limits of the_last two aberra-tions is discussed.

-Jour. App. Phys. Vol. 13, No. 9(1942), page 582.

The Theory of Units(L. H. Bedford)

In the formulation of electromag-netic theory there are a three stagesat which arbitrary constants are con-veniently introduced for the purpose -of defining units. It is customary toassign immediately the value unity tocertain of these constants, after whichthey are lost sight of. In the presenttreatment, a sketch of electromag-netic theory is given in which the sup-pressiOn of these fundamental icon-stants k,, k,) does not 'occur.Maxwell's theory is shown to lead to acertain relationship between the k'sinvolving the velocity of light. Sub-ject to this restriction, the assignmentof k -values is an arbitrary matter, theprocess of formulating a unit systembeing one of 2 degrees of freedom.

A table is constructed showing theassignment of k -values correspondingto the various known unit systems.This table can, amongst other things,be used to derive the numerical rela-tionship between the unit quantities ofthe various systems.

Serious inconsistencies of methodand nomenclature in connexion withthe practical system of units arebrought to light and proposals forrationalisation are considered.

-Jour. Brit. I.R.E. to be published.



THE world-wide use of " AVO '1 Instrumentsis striking testimony to their outstandingversatility, precision and reliability. In everysphere of electrical test work they are appreciatedfor their dependable accuracy, which is often

used as a standard by which other instrumentsare judged. There is an "AVO " Instrument forevery essential electrical test.

The Model 7 Universal Avometer (illustrated) is a self-contained combination A.C./D.C. meter providing for 46ranges of direct readings of A.C. and D.C. voltage ; A.G. andD.C. amperes; resistance; capacity; audio -frequency power out -put and decibels. Any range instantly selected by means of tworotary switches. No external shunts or multipliers. 5 -in.

hand -calibrated scale with anti -parallax mirror. Dead beatin action. Current consumption negligible. Conforms withB.S. 1st Grade accuracy requirements. Automatic cut-outdisconnects meter from supply in event of severe overload.Automatic compensation for temperature variations. Robustlybuilt to ensuring lasting accuracy.

Regd. Trade Mark

ELECTRICALMEASURINGINSTRUMENTS

Some delay in delivery of Trade orders is inevitable,but we shall continue to do our best to fulfil your

requirements as promptly as possible.

Sole Proprietors and Manufacturers :

THE AUTOMATIC COIL WINDER & ELECTRICAL EQUIPMENT Co. Ltd.Winder House, Douglas Street, London, S.W.I Phone : VICtoria 3404-7

We are pioneers in the manufac-ture of extruded plastics and ourspecialised knowledge is entirely atyour service. If you believe extruded

plastics will improve your productswe shall be pleased to advise you.

TENAPLAS LTD. Head Office 7 PARK LANE, LONDON,W.I

THE " FLUXITEQUINS " AT WORK.

"All right m'lads,

But I'm what they callversatile,'

I can work dayand night

Sold'ring wireswith FLUXITE

But weather like thiscramps my style!"

See that FLUXITE is always by you -1n the house-garage-work-shop wh speedy soldering is needed. Used for 30 years ingovernment works and by leading engineers and manufacturers.Of ironmongers-In tins, 8d., 1/4 and 2/8. Ask to see the FLUXITESMALL -SPACE SOLDERING SET-compact but substantial-complete with full instructions, 7/6. Write for Book on theart of "soft "soldering and ask for Leaflet on CASE -HARDENINGSTEEL and TEMPERING TOOLS with FLU XITE. Price Id. each.

ITo CYCLISTS I Your wheels will NOT keep round and true unless the spokesore tied with fine wire at the crossings and SOLDERED. This makes a muchstronger wheel. It's simple-with FLUXITE-but IMPORTANT.

THE FLUXITE GUN puts FLUXITEwhere you want it by a simple pressure.

Price 1/6, or filled 2/6.

FLUXITE LTD. (DEPT. E.E.),BERMONDSEY ST., S.E.I.

ALL MECHANICS WILL HAVE

FLUXITEIT SIMPLIFIES ALL SOLDERING



PATENTS RECORDthe information and illustrations on this page are given with the permission of the Con-troller of H.M. Stationery Office. Complete copies of the Specifications can be obtained

from the Patent Office, 25, Southampton Buildings, London, W.C.2, price Is. each.

INDUSTRYElectric Motor -Control Systems

Relates to control systems for con-trolling the speed of an electric motorwith respect to a reference ,speed,especially for machines operating ona length of material such as a web ofwood veneer, paper or cloth,

In veneer stripping or cuttingmachines in which the veneer isstripped from a log and run out on amotor -driven roll conveyor the dia-meter of the log decreases as theveneer is stripped off. Therefore thelinear speed at take -off decreases con-tinuously even though the log isrotated at a constant speed. If theconveyor continues to run at a givenspeed undue tensioning and perhapsbreakage will occur.

In the application of the inventionto veneer stripping apparatus, thefirst of the pilot generators ismechanically coupled to the web as itleaves the log, i.e., at the point wherethe linear speed of the web is decreas-ing. The second pilot generator iscoupled to the driving roll which de-termines the speed of the web on theconveyor. The pilot generators areconnected to a common output ,circuitso that the respective output voltagesoppose each other. If the voltagesvary with respect to each other becauseof a change in speed of one or theother of the generators, a circulatingcurrent, which is proportional to thespeed difference, flows in the saturat-ing winding of a reactor.

The resulting change in impedanceincreases or decreases the output ofvalve circuit to correspondingly in-crease or decrease the field excitationof the main generator supplying thedriving motor. Thus, the speed of theconveyor motor is raised or loweredto give a uniform linear speed of theweb.

A simple anti -hunting circuit is pro-vided to prevent over -correction ofthe generator voltage output and driv-ing speed.

-The British Thomson -HoustonCo., Ltd. Patent No. 540,035.

CATHODE-RAY TUBESImproved Electron Camera

To overcome the disadvantages of adiscontinuous (mosaic type) photo -emissive electrode and to avoid thedistortion introduced by a double -sided mosaic it is proposed to make anelectron camera in the form of a tube

with a semi -circular extension tube,at the far end of which is mounted aflat semi -transparent photo -cathode on'which the optical image is focused.

At the end of the curved tube near-est the electron gun is a target elec-trode of mosaic structure offset fromthe axis of the beam from the electrongun.

A coil surrounds the whole en-velope, the toroidal portion serving toguide the photo -electrons round thecurved portion of the tube on to themosaic, where they produce an elec-tron image.

Electrons from the gun pass throughan aperture at the side of the mosaicplate and passing round the curvetube, are reflected back from thephoto -cathode which is at a high nega-tive potential to the source of elec-trons. These electrons are thencaused to scan the mosaic on the sameside as that on which the electrostaticimage is formed and so discharge themosaic in the usual way.

-Marconi's W.T. Co. and H. A.lams. Patent No. 546,559.

Improved Cathode Ray Tube

To provide a cathode-ray tube ofsimplified and improved constructionespecially suitable as the image repro-ducer in television systems.

This comprises a thermioniccathode, a first electron lens consist-ing of a cylindrical apertured gridand a cylindrical first anode. Thegrid has an external diameter lessthan 0.15 times that of the internaldiameter of the anode. The gridcathode are preferably as small asmanufacturing facilities permit, thecathode electron emission being atleast as high as so milliamperes persquare centimetre of emissive surface.The decrease in the area of the elec-tron -emissive surface of the cathoderesults in a substantial reduction ofthe heater wattage.

An additional advantage is that thegrid to cathode capacity is materiallydecreased by the relatively smalldimensions of the electron gun com-ponents, and is claimed to be of theorder of one or two opF.

-Hazeltine Corp. (Assignees ofR. C. Hergenrother). PatentNo. 546,846.

December MeetingsI.E.E. Wireless Section

On Wednesday, December 2, at 5.30p.m. in the Lecture Theatre of the In-stitution of Electrical Engineers,Savoy Place, W.C.2; a paper on "TheElectrical Amplifying Stethoscope andPhono-electrocardioscope" will beread by Dr. G. E. Donovan,M. Sc., M. B ., B. Ch .

Institute of Physics-" NewtonTercentenary "

A meeting of the Branch will beheld in the Lecture Theatre of theRoyal Institution, Albemarle Street,Piccadilly, London, W. i, on, Wednes-day, December 9, at 5.o p.m., to cele-brate the three -hundredth anniversaryof the birth of Sir Isaac Newton.Papers will be read as follows :-

(a) " Life of Newton," by H.Lowery, D.Sc., F.Inst.P.

(b) " Some Historical Aspects ofNewton's Published Works," byH. F. Buckley, D.Sc., F.Inst.P.(National Physical Laboratory).

This meeting, which is of generalinterest, will be open to visitors. '

The Television SocietyThe annual general meeting of the

Television Society will be held onSaturday, December 5 at 2.30 p.m. inthe small theatre of the Institution ofElectrical Engineers, -Savoy Place,W.C.2.

At the conclusion of the meeting(which is for members only) a shortreview of " Progress in Colour Tele-vision " will be given by the Hon.Lecture Secretary, Mr. G. Parr.

An informal reunion will be held atthe conclusion of the meeting, whichnon-members may attend by invita-tion only.

I.E.E. Students SectionAt the next meeting to be held on

December 7 at 7 p.m., a paper willbe read by Dr. D. S. Anderson on" The Failure of the Technician inhis Role of Citizen."

Tea will be served from 6.30 p.m.On December 12 a dance will be

held at the Lysbeth Hall, Soho Square,W.1, from 3-7 p.m.

Tickets, price 2s. 6d. single, 4s. 6d.double, can be obtained on applica-tion to Mr. A. N. Lord, 8 HaddenWay, Greenford, Middlesex.



WAIT FOR IT !-and then when normal conditions arerestored, Reliance Potentiometers, in-corporating our many years of experience,will again be available.

Until then National requirements mustcome first. If we can help you in thisconnection, we will. Therefore do nothesitate to write us.

RELIANCEManufacturing Co. (Southwark) Ltd.Sutherland Road, Higham Hill,Walthamstow. E. 17.

Telephone-Larkswood 3245

4131J%&92/6ipmeneIllustrated above :--30 wattAmplifier with Radio

Available for Priority Work.

TR' X ELECTRICAL CO., LIMITED6S, Bolsover Street, London, W.I.'phone: EUS 5471/2. Grams: Trixadio, Wesdo, London.

Announcement

We manufacture Low Loss

Ceramic Parts at least asgood as any on the market

but we regret that we areat present unable to accept

orders for these as our pro-

duction of these materials

is already being used forwork of national importance

A

T Lik 1

?4a/'TT/44

Head Office: Eastwood, Hanley,Staffs. London: 85, StreathamHill, S. W.2. Factories at Han-ley, Stone and Lonyton, Staffs.Telephone : Tulse Hill 5255-6

and Stoke-on-Trent 5272-4.

THE SUPERHETERODYNERECEIVER

By Alfred T. Witts, A.M.I.E.E. A reliable, compre-hensive and up-to-date guide to the superheterodynereceiver, incorporating all recent developments.Fifth Edition. 5s. net. " Gives all the informationnecessary for a complete understanding of thesuperheterodyne receiver."-Practical Wireless.

PROBLEMS IN RADIOENGINEERING

By E. T. A. Rapson, A.C.G.I., etc. A classifiedcollection of examination questions set from time totime by some of the more important examiningbodies in Radio Communication, together with someuseful notes and formulae. Fifth Edition. 5s. net.

CATHODE RAY TUBESBy Manfred Von Ardenne. In this book a very widefield is covered, and the book deals with fundamentalprinciples and early developments as well as withpresent day methods and apparatus. 42s. net.

PITMAN'S FOR RADIO BOOKSThere are over thirty books in our radio list. Writeor call for a copy.

PITMAN 39 PARKER STREETKINGWAY, W.C.2

====The Golden

FleeceWhen Jason set out with his Argosy insearch of the Golden Fleece he did so withno greater determination than we atGardners when we embarked upon thebuilding of a range of quality transformersin our search for perfection. Although everyGardner product bears the stamp of thisdetermination it is prominent indeed in ourSmall Power Transformers up to 4 kva.When next you need this type of compon-ent and a " Golden Fleece " quality is a

must ' let your Argosy set sail in ourdirection.

We regret that at present Small Power Trans-formers are available for highest priorityorders only.

QUALITY MAINS"1111111111MINIMMIMMOMS.

TIIANsFoRMElls

GARDNER'S RADIO, LIMITED,SOMERFORD CHRISTCHURCH HANTS.



A Pulse Generator forCircuit Testing

AUSEFUL pulse generator cir-cuit is described by H. P.Manning and V. J. Young in

the May issue of The Review of Scien-tific Instruments. It produces a pulsewhich may be conveniently varied induration, voltage and frequency so asto simulate actual operating condi-tions in a Geiger counter or similarapparatus. It is also useful forchecking amplifier response.

C, and R, make possible a continu-ous frequency range from o.8 to 30,000pulses per sec. C2 allows a choice ofpulse brea.dihs varying from 4o micro-seconds to 3 milliseconds. R, variesthe pulse voltage from zero to i5ovolts. The reversing switch changesthe polarity of the pulse withoutchanging its shape or disturbing theground connexions. All five controlsare essentially independent of eachother. The valves are (from left toright) 885; 6C6; 42; 42.

C6

OutputT

Values of Components.R1

12,

R3

R,R5

1-megohm potentiometer5000 -ohm potentiometer500,000 ohms15,000 ohms1,000 ohms

{2 µf1µf0.4

gf0.15 µfC,0.065 pl 9 -position

R,R,R8

50,000 ohms10,000 ohms100,000 ohms

0.03 kif0.01 ill0.003 i.il

switch

R, I megohm 0.0007 /IfR10 100,000 ohms /../f

R11 100,000 ohms 1450 AufR12 500,000 ohms C2 500 gp.t. 5 -positionR13 25,000 ohms 250 µpi switchR14

R15

50,000 ohms600,000 ohms C3

< 100 ILO0.001 /If

R16 12,500 ohms C4 = C5, 0.5 [ifR17 I megohm C2 = 0.05 pl

the

STRATOSII:series of

SEALEVI 11 RA 0 S

111

for use at all altitudes

1

2

3

All steel construction-even tothe rivets-ensuring uniformexpansion under extremes oftemperature.

Reed driving coil-wound on abakelite moulded bobbin to meetall climatic conditions.

Metal can, sponge rubber lined-Acoustically and electricallyshielding the Vibrator.

4 Driving contact of non -tarnishableprecious metal-ensuring startingunder the lightest of pressuresand voltages.

5 Contacts ground almost to opticallimits.

6

7

Stack assembly. Mica and steelonly are used.

International Octal base sealedby the WEARITE STRATOSILprocess.

Embodying " Oak Manufacturing Co.'s "patents 460470, etc.

WRIGHT & WEAIRE LTD.HIGH ROAD, TOTTENHAM, N.17. Telephones: TOTtenham 3847-8-9

2

3

4

5

7


December, 1942 Electronic Engineering III

t ARDux,is a quick setting

adhesive for mouldings and laminatedplastics. No preparation of the surfacesis necessary. ' Ardux ' is doing a valu-able job in eliminating wastage and inreducing machining in the productionof aircraft components. ' Ardux ' canprobably help you in your problem;

May we show you how ?

Aero.IResearch Ltd., Duxford, Cambridge.Telephone: Sawston 167-8

LOUD

SPEAKERSTHE [WORLD'S FINEST REPRODUCERS

TRANSFORMER

LAMINATIONSCore Widths i1," to II" (E's and I's.)

EIGHT STOCK SIZESA Comprehensive Bulletin together with details of Associated Coversand Clamps with design data will be sent to manufacturers on request.

BRITISH ROLA LIMITEDMINERVA ROAD PARK ROYAL N.W.10 WILleaden 4322

"AFTER THE WAR" TELEVISIONWith the coming of Peace, and backed by the Technicaladvances that have resulted from the war years, TELE-VISION will again spring to life and offer opportunities,as never before, to the trained man.Now is the time to prepare for this new " market." OurHome -study " Television " Course is comprehensive,up-to-date and suitable both for the amateur and ex-perienced Radio Engineer.

We definitely guarantee"SATISFACTION-OR REFUND OF FEE"

You are advised to send for details of this special Television Courseimmediately, to Investigate its possibilities. Details will also gladly besent of A.M.I.E.E. and A.M.Brit.I.R.E., Examinations and othercourses In all branches of WIRELESS ENGINEERING, includingparticulars of our special SHORT WAVE and RADIO SERVIC-ING C OURSES which are in such demand-Free & post free.

BRITISH INSTITUTE OF ENGINEERING TECHNOLOGYPrincipal : Prof. A. M. Low.

337, Shakespeare House, 17/19, Stratford Place, London, W.I.

A-ex-is-Z(7nteismgwcak

SINEW WORKSR EDDITCHTEL.10 0 NGLAND ESTAB.1920

offallitok/gob',, ,

\\NRA1111411111

AdI II,

WATERPROOFU NO i " U NO

41TENPRooF STENCIL 04DRAWING INKS.'ATENTNO, 27170-53

1

'l are suitable forBLACK 1 every drawingoffice use.

A. WEST & PARTNERS, LTD. 36 BROADWAY, WESTMINSTER, S.W.1


iv Electronic Engineering December, 1942

WITH BANDPASS CRYSTAL FILTER

The advantage crystal control gives in

improved rejection of interference outside

the band and correspondingly better

signal-to-noise ratio has been fully ex-

ploited by Eddystone designers. This

receiver may be inspected at 14 Soho

Street, London. Supplies limited

WEBB'SRADIO

to The Short -Wave Specialists

OFFICIAL AUTHORIZATION ONLY HOURS OF BUSINESS : 9 A.M. TO 4 P.M. SATS. 9 A.M.-12 NOON

14 SOHO STREET, LONDON, W.I. Te% Gerrard 2089

Printed in Great Britain by The Press at Coombelands, Ltd., Addlestone, Surrey, for the Proprietors and Publishers, Hulton Press, Ltd., 43-44 Shoe Lane, London. E.C.4,Sole Agents for Australia and New Zealand : Gordon and Gotch (Alsia), Ltd. South Africa : Central News Agency, Ltd.

Registered for Transmission by Canadian Magazine Post.


engineering - worldradiohistory.com...wave- analysis-part 2 a simple electronic switch electron...

Documents