Vol. 9, No. 9 '"2óï

SMALL SELENIUM RECTIFIERS

by J. J. A. PLOOS van AMSTEL.

:A. selenium rectifier consists essentially of a layer' of selenium, a blocking layer anda layerof metal acting "as cathode. The properties of -such a' rectifier depend to a large extentupon the reaction of the latter metal when the selenium is applied. Some metals uscdas cathode give rcctifiers with an' exceptionally low resistance in the transmitting directionwhilst other metals make the rectifier suitable for rectifying relatively high voltages.This rectifying is further promotea "by applying all additional or ffi:iificial blocking 'layer.With thesé methods Philips have developed three kinds or"selenium rectifiers, two 6"fwhich are made ëxchisively in small ~izes (a few milliméters) and the third also in largerdimensions. :Here ,the main properties- and applications of the' small rectifiers will bediscussed.

In tp.e course of time blocking-layer rectifiershave "been developed into electrical switching

. elements with a great variety of properties, dimen-sions and uses. An article dealing with the -generalcomposition of a blocking-layer rectifier and thétheory of its working was published in this journalin 1939 1), followed a year later by another, article 'on their uses and in particular the application ofselenium cells in rectifiers 2). Here we shall gosomewhat farther into some details of the construc-tion of these selenium rectifiers. It will he foundthat the properties .of the final product "can beinfluenced in à certain direction by the choice ofce~t~mmaterials and their ~ethod of treatment ..,In this-way it is possible' to 'attain the best solutio~for a given object. '- As a rule a rectifier is required to have the lowest'possible resistance in the transmitting direction, thehighest' possible resistance in the blocking directionand the least possible capacity; These requirementsare partly 'contrary to each other, for a rednetion ofthe resistance in the transmitting direction, for in-stance, is accompanied byreduction ofthe resistancealso in the blocking «¥rection. The choice of prefer-ence depends 'entirely upon which property is to bestressed for the application in view. In the following~t will he' shown that means are availahle' to attainthis end. We shall consider here in particular smallselenium rectifiers (the largest size is only à few mm)because they have the most varied· applications.

Let us first recall the general composition of àselenium rectifier (fig. la): hetween ~ layer of semi-conducting selenium and a layer of good conductingmetal is a very thin insulating blocking layer.Electrons appear to pass more easily. into theblocking' layer from the" metal. "rich' .in free"

1) w. ca. van Geel, Blocking-layer rectifiers, PhilipsTechn. Rev. 4, 104-110, 1939.

2) D. M. Duinker,The use of selenium cells in rectifiers,Philips Techn. Rev; 5, 199-207, 1940, '

621.314.634

"

"

eléctrons than 'from the selenium, which' is "poor in:electrons. Thus 'a positive ' current is' more readilybrought about in the direction of the- arrow (trans-_mitting directiori] than the other way (blocking'direction]. ,~.

- .. po.

tJ2.

" 50Q47

Fig.!. a) Diagrammatic representation of a selenium rectifier,P = meta]. carrier plate, Se = layer of semi-conducting sele-nium, S =' blocking layer, M = layer of good conductingmetal or alloy. ' '. " .. ; , .'.' b) ,For comparison, .a.valve with hot cathode, Theemitting. cathode K corresponds to the metal M Iikewise emitting,electrons; '--:.

If a selenium rectifier through which ëurrent flowsin- the transmitting' direction is compared with a.hot-êathode valve (fig. 1b) one may say - withoutgoing into the difference in mechanism - that the

. metal oorresponds to the hot cathode (hence the. '. ./term cathode metal or cathode layer). In manycases it is well worth bearing this in mind.

Influenee of the c~mpositi~ns of a selenium re~tifierupon its proJ_lerties',

. -A selenium rectifier may be made.in the following,

way: molten selenium is poured onto a metal carrierplate, serving to lend strength to the whole, thenpressed flat and, subjected to a heat treatment,thereby forming' on, the surface of the selenium aninsulating 'Iayer called the .génetic or n a tur a lblocking layer (in contrast' to the artificialblocking layer which will he mentioned farther on).Finally a layer of good conducting metal (maybean alloy) is added in some way or other, for instance

\ '

268 PHILIPS TECHNICAL HEVIEW 1947/1948

, by extrusion, by vaporization in vacuo or by atomi-zation in a gas discharge.

both directions.This is utilised in one of the kindsof rectifiers manufactured by Philips,

Artificial blocking layer

Instead of leaving it at a genetic blockinglayer.extended or not with insulating Se-metal com-pounds, before precipitating the cathode layer onthe genetic blocking layer an artificial blockinglayer can be applied to the latter. Various substan-ces can he chosenfor this (especiallycertain organiccompounds). Such an artificial blocking layer,which is alwaysrelatively thick, checks considerablythe reactions mentioned above, so much that inas far as they take place at all' they have littleeffect. The capacity between the electrodes of sucha rectifier,per surface unit, isright at the outset con-siderably lower than that of a rectifier withoutartificial blocking layer, but at the same time theinternal resistances III both directions are muchhigher.

Reactions between cathode metal and selenium

, The rectifying properties of a rectifier made inthis way appear' in the current-voltage charac-'teristics, which can be recorded in both directions(how this can best be done will he shown later).Another property of great importance in someapplications is the capacity that can he measuredbetween the carrier plate and the cathode layer,these forming the external ele'ctrodes. It .appearsthat both 'the magnitude of this capacity and thetrend of the current-voltage characteristics dependto a high degreeupon the nature of the metal or the 'alloyfrom whichthe cathode layer is made and alsoupon the treatment of the rectifier. In the main theresult is determined by the reactions taking placebetween the cathode metal and the selenium. Twocases are to be distinguished; that where the reac-tion between the selenium and the metal formsnon-conducting s_phstancesand that where the Methods of ,measuringreaction products a~e indeed conductive. Generally Beforepreceding to show~hat types of seleniumsuch reactions can be promoted by heating the rectifiers Philips are making according to theserectifier to a high temperature. Furthermore, the aspects, 'wewill briefly deal with the methods ofreactions forming'non-conducting compounds can measuring.. be promoted' by heating to beyond the meltingpoint of the cathode metal and at the same time Recording the dynamic current-voltage characteristicsapplying-a voltage making the metal positive with The characteristic (current as function ofvoltage,respect to the selenium. This causespositive metal ' either in the transmitting direction or in the block-ions to pass over to the selenium. ing direction)can he recordedby varying an applied.Where we have .to do with insulating (or at direct voltage step for step and measuring the

least very poorly conducting) Se-metal compounds current at each step, In the article quoted in foot-these may be regarded as an accretionofthe block- note 2), however; it was stated that a staticing layer. ~en a rectifier in 'whichsuch compounds characteristic derived in this manner will as a rulemay arise is subjected to heat 'treatment one may .differ from' the relation existing between simul-observe a rednetion of the conductivity in both taneous momentary values of current and voltagedirections and also of the capacity, corresponding when the rectifier is subjected to a' rapidly alter-to the picture of a thickened blockinglayer. Exam- nating voltage (dynamic characteristic). Since'pIes of metals'forming with selenium poorly con- this is the case in hy far the majority of the appli-.ducting compounds are. cadmium, magnesium and cations of these rectifiers, we are in the first place .aluminium. We will come back to the action of interested in the dynamic characteristics. Thesecadmium' presently. characteristics can be made visible by' means of aIf,on the' other hand, we use in the cathode layer cathode-ray oscillograph. Fig. 2 shows the circuit-

metals which with selenium form highly con- ing arrangement for recording these dynamicduc t ive compounds ...:_examples of such metals characteristics.'are. gold, silver and antimony - then" the layer'forming these compounds may he regarded as acontinuation of either the cathode layer or theselenium.The blocking layer itself may in the firstinstance remain unchanged, Nevertheless, also in\ this case it will as' a tule 'be observed that thecharacteristics have undergone"changes,during thereaction, often'in the sen~eof a lower resistance in

Measuring the capacity

Capacity can be determined with the aid of thebridge circuit illustrated ui fig, 3. A variable directvoltage E is laid on, the rectifier in the blockingdirection and on that 'a smallaltèrnating voltage ofsay 10mVis superposed. Sincethe measuringresultdepends somewhat upon the frequency it is prefer-

-I·"

Vol. 9, ~o. 9-

. SMALL SELENIUM RECTIFIERS _269

A

a

b

50Q48

Fig.2. a) Circuiting schemes for recording dynamic charac-teristics of blocking-layer. rectifiers: (a) in the transmittingdirection, (b) in the hlocking direction. Since different scalesare desired for the transmitting and the blocking character-istics, the recordings are taken from different circuits. In (a)the resistance RI is small compared with the resistance in thetransmitting direction of the valve G; the voltage VI (about· •1 V) is therefore practically equal to the voltage on the rec-tifier. The higher voltage V2, supplied by the same transformcrT as gives VI' is thus at thc same time a measure for the vol-tage on the rectifiers; it supplies the horizontal deflection onthe oscillograph tuhe O. The voltage at RI' which is a measurefor the current passing through the rectifier, is conducted viathe amplifier A to the plates for the vertical deflection.

b) The valve G carries the sum of the alternating voltageV3 and a direct voltage from the capacitor C. The leakagecurrent passing through the rectifier when Va has the givenpolarity causes'in the resistor R2 a voltage loss which suppliesviet the amplifier A the vertical deflection on the oscillographtube 0, whilst the voltage on G provides for the horizontaldeflection direct.

able to carry out this measurement 'with a fre-quency within the range in which the rectifier is. tohe used (in the case of certain modulator cells forinstance 60 kcfs). The bridge is balanced as far aspossible by adjusting the resistor R and the capacitorC. ~he value read for C is then the capacity re-

50Q4Q

Fig: 3. Bridge circuit for measuring the capacity of blocking-layer rectifiers. Tl = transformer supplying on the secondaryside a voltage of 2 X abt. 10 mV with a frequency of say -60 kc/s; G = valve to be tested; R and C d variable resistorand capacitor for bringing the bridge into equilibrium; T2 =output transformer connected to an amplifier A and a record-ing instrument I; E = variable direct voltage, upon. whichthe capacity found grcatly depends. The capacitor in serieswith the secondary winding of Tl. prevents direct current• flowing through this coil.

quired. This depends for a great deal upon the directvoltage E, which fact must be taken into account.when compa~g one rectifier with another. :

For the explanation of the fU:ctthat the capacity dependsupon the" voltage, we refer to what is stated in the' articlequoted in footnote 1) about insulators and semi-conductors.In an insulator each electron is bound to its place (i.e. a certainatom) whereas in a semi-conductor the electrons have freedomof movement, due' to the iact that there is either an excess ofelectrons or else an electron is lacking in places. In the lattercase conduction is brought about by the movement of. elec-trons from one open place to the next. If all the open placeswere occupied by electrons there would be no conductivity.One can imagine something similar happening in the sele-

nium when ä direct voltage is connected to a selenium rectifier(Se negative, metal positive, fig. 4): electrons in the selenium

Fig. 4. The voltage E in the circuit of fig. 3 causes charges oneither side of the blocking layer S, positive in thc metal at Mand negative in the semi-conducting selenium Se. This givesrise to an area Goin the latter layer where all open places areoccupied by electrons and which consequently rio longer hasany conductivity, This area G becomes thickerthe higher thevoltage E, which accounts for the drop then ·found in thecapacity.

"are th~n attracted in the direction of the blocking layer; thethin layer of the .semi-conductor immediately. adjacent tothe blocking layer thus becomes saturated with electrons.The higher the voltage applied, the farther this chargeextends into the semiconductor, causing a corresponding drop.in thè capacit~ between the tw~ elect~odes.

A discussion' of three kinds. of selenium rectifiers

The demand in various practical fields-for block-ing layer rectifiers with different properties has ledto the manufacture of three kinds, which we willrefer to in this article as I, H and Ill, each of whichis made in the dimensions most suitable for a certainpurpose. Per unit of effective surfacè the capacity'decreases and the resistance in the transmittingdirection increases in the order of I-II-HI. Theserectifiers will be discussed in the order of H-III-I.

,I

Type II

The oldest is type II, in which an alloy of tin,cadmium and bismuth is applied direct to the ge-netic blocking layer. As already remarked, togetherwith selenium the c~dmium in this alloy .forms aninsulating compound ~hich can be considered to

270 PHJLIPS TECHNICAL REVIEW j 947/19~1l

a

b

-C,ó' -0,4 0 o,4 V O.:Ju'd9

c

Fig. 5. Dynamic characteristics for the transmitting directionderived with the circuiting schemeof fig. 2a and, as in the caseof the characteristics shown in figs. 6 and 7, recorded with anoscillograph of the type GM3156. Amplitude of Vl: 0.68 V.a) CeU of type I, diameter of cathode layer 1.5 mm.b) CeU of type 11, diameter of cathode layer 3 mm.e) Cell of t}'Pe Ill, diameter of cathode layer 3mm.The left-hand current scale in (a) indicates the currents mea-sured, whilst the right-hand scale, which would apply for adiameter of 3 mm, is added to facilitate comparisonwith the4 times as large cells of (b) and (c). To the left of the i-axis isto be seen a part of the blocking characteristic, which practi-cally coincideswith the negative axis of abscissa.

belong to the blocking layer. Such a compound hasa favourable action owing to the raising of the re-sistance in the blocking direction and reduction ofthe capacity; against this, however, is a highertransmitting resistance.

The reason why cadmium is used in combinationwith tin and bismuth is that with these three metalsan alloy can be made which has a low melting point,so that it is easy to spray, and which can be kept inliquid form without oxidising too much. Further-

o

more, this alloy can be used for soldering one of theconnecting wires,Dynamic transmitting characteristics of a rec-

tifier of the type II are given in figs Sb and 6brespectively for 0.68 V and 1.34 V amplitude of thevoltage VI (cf. fig. 2a). The dynamic blocking cha-racteristic is represented in fig. 7b. lnfig. 8 curve 11shows the trend of the capacity as function of thedirect voltage applied E. We shall revert to thesefigures when discussing types III and I.

Type lIJIn the rectifiers of type III the genetic blocking

layer is reinforced with an artificial blocking layer

a

b

c

Fig. 6. The same as fig. 5 but with a voltage amplitude of1.31V.

Vol. I), No. 9 51\1.·\u. SELENIUM RECTIFIERS

a

b

c

Fig. 7. Dynamic characteristics for the blocking directionmeasured on the three cells I, II and III in the circuitingarrangement of fig. 2b. In this case the cells were heavilyoverloaded as regards voltage. It appears that the blockingcharacteristic often shows a peculiar loop shape, as is parti-cularly the case with cell II; the direction of flow through theloop is indicated by the arrow. The maximum blocking vol-tage permissible under normal working is indicated by avertical stroke. To the right of the axis of ordinates a part ofthe transmitting characteristic is to be seen.

before applying the Se-Cd-Bi alloy. This improves,in the first place, the blocking properties, as mayappear when comparing fig: 7e with fig. 7b; both ofwhich are the characteristics of rectifiers having7 mmê effective surface. It is seen that with type IIIthe voltage can be raised much higher without caus-ing a rapid increase in the leakance (which may beregarded as an indication of breakdown). This isparticularly of importance for the rectification of

~71

high voltages, for when using the type III rectifiersa smaller 'number in series suffices.As was to be expected, with the improved block-

ing properties one has to take into the bargain aless satisfactory transmitting characteristic: the"threshold voltage" (at which the current becómesnoticeable) is, it is true, only little higher in type IIIthan in type Il (compare fig. Se with fig. Sb measuredwith rectifiers of the same size and represented onthe same scale), but the slope of the reat of thecharacteristic differs considerably (compare fig. 6cwith fig. 6b, wherethe scales of current are not equal).In the case of the type III rectifiers one will also

expect a smaller capacity than that of type Il. Thisis confirmed by the measurements (compare curveIII with IJ in fig. 8).Rectifiers of the type III have been standardised

in a series of sizes from 7 mmê up to 14 000 mmf

effective surface 3).

Type l

Last of all we have type I, which in its construc-tion and in its properties differs appreciably from

IIIIIIIIIIII

J'I'///

/

",,"'/'

3000pF

C

r2500

2000

1500

o +1 +2 V---E

50951

Fig. 8. The capacity C measured with the circuiting schemeof fig. 3 plotted as a function of the applied direct voltage E,for the rectifiers I, IJ and I I I. The curve I has been derived bymultiplying the ordinates of I by 4.

3) The selenium cells dealt with in the article quoted in foot-note 2 are of type TU .

•

( ...

~72 PHILIPS "I:ECHNICAL REVIEW. 1947/1948

types 11and Ill. As a matter of fact it was developedfor quite ,a different purpose, viz. to ,get the most :favourable characteristic possible in the transmirtingdireetion. The use of the cathode metal that forms aconductive compound with selenium has been mostsuccesful. Since, however, the melting point' of themetal in. question is much higher than that of sele-nium, the cathode layer cannot be àpplied in liquidform, as is the case with the alloy previously men-tioned. It was found possible to' precipitate a thinlayer of the metal by atomization, Dut then theproblem arose how to make a good electric contactwith it. A contact spring is not reliable in the longrun. Soldering on such a very thin layer is out of thequestion, whilst furthermore soldering over the me-.tal would lead to reactions which are undesirable in. this case, because they would result in the favourableproperties in the transmitting direction being lost.

50952

Fig. 9. Cross section of a cellof type I.P = carrier plate, Se =selenium layer, S = ,genetic blócking layer, 1I-i =' cathodeobtained by vaporization, L == lacquer, Leg = alloy for solder-ing the inlet lead. The lacquer allows enough of tbe alloy topass- through to ensure a good electric contact; the obnoxiousCd-Se compound is formed on only a very small part of theeffective surface.

This difficulty was solved' in the manner" sketched. in fig. 9. After a layer of metal with a diameterof about 1.5 mm is applied by vaporization (thereason why this diameter has been èhosen so smallwill be clear farther on) the whole plate (diameter6 mm) is coated with a thin layer of a suitablelacquer. Then a layer of the aforementioned Se-Cd-Bi alloy is applied in the middle, with which thelead wire can be soldered. Enough of the alloypenetrates through the pores of the lacquer to makea reliable electric contact in a number of places withthe first layer of metal. At these places some reac-tion will take place between the selenium and thesolder through the very thin cathode layer (formingthe undesirable Cd-Se compound), but this takesplace on such a small fraction of the effective sur-face as to have scarcely any effect upon the charac-teristic of the rectifier. . .

•

The characteristics measured for' such a rectifiertor the transmitting direction are to be seen in figs.5a and 6a (the small current scale applies for a cellwith 1.5 m~ diameter of the cathode layer withwhich the measurements were carried out, whilstthe large current scale applies for a surface 4 timesas large and has been added to facilitate a compa-.rison 'with the figures 5b and c and 6b and c, relatingto rectifiers with a diameter of 3 mm). Itis seen thatthe threshold voltage is less than half that of ~herectifiers of types 11and III and that the rest of thecharacteristic has a much steeper slope.Fig. 7a shows that the rectifiers of type I can

bear less voltage in the blocking direction than, theother types, ~but this is of no importance in appli-cations where the voltage on the rectifier does notgo beyond a few volts. What is more serious is thatthe capacity (taken for the same surface) is much

. higher than in the case of types 11 and III (seecurve l' in fig. 8);As a matter 'of fact this is one ofthe reasons why the diameter 'has been reducedfrom 3 mm to 1.5-1.2 mm; the capacity is then atleast 4 times as small' (curve I, fig: 8) 4). Since with a,given current intensity the density of current in thetransmitting direction is then 4 times as great, therectifier works in a less curved part of the charac-teristic, which for some applications is a greatadvantage: the power involved is so 'small that the .heat dissipation --in large rectifiers one of the mainfactors restricting the admissible cur~ent density -becomes quite negligible.

Fig. 10 shows some seleniu?l rectifiers in differentstages of production and mounted in various ways.

Applications of small selenium rectifiers

The most, important applications of small sele-nium rectifiers may be classified under threeheadings:1) in electrical measuring instruments,2) in modulators in carrier-wave apparatus,3) in "small rectifiers".

1) Selenium rectifiers in electrical measuring in-.struments

The uses to which these rectifiers are put inmeasuring instruments lie in quite different fields. Inthe first place blocking layer rectifiers are oftenused for converting an alternating current that isto be measured into a pulsating direct current, the

4) The thickness of the layer of lacquer compared with the.blocking layer is such that the capacity between the pro-truding edge of tbe alloy and the selenium can be ignoredcompared with the capacity between the cathode layerand the selenium .

Vol. 9, No. 9 SMALL SELENIUM RECTIFIERS

average value of which is measured with the aid ofa moving-coil instrument. The main advantagesof this method are that the meter uses very littlecurrent (as compared with most other measuringmethods) whilst the scale is more or less linear.Another field of application comprises the use of ablocking-layer rectifier as shunt for a measuringinstrument, for instance to guard it against over-

If the rectifiers were free of capacity than themeter reading would be independent of the fre-quency. The measuring instrument would thenindicate the average value of the alternating current10, given by I = Vj(R + T), where Vrepresents thealternating voltage to be measured, R the seriesresistance and T the sum of the resistance Tm of theinstrument and twice the transmitting resistance ï d

Fig. 10. Kinds of selenium rectifiers, some mounted and some not. From left to right:(front row) five cells of type I in different stages of construction (without cathode layer,with cathode layer, with lacquer and alloy, with inlet leads) and three square cells oftype Ill;(second row) four rectifying cells destined for a voltmeter in Graetz circuit housed ina box of "Philite"; the box filledwith insulating material; the lid of the box; the completerectifier;(third row) hermetically sealed ceramic tube containing a modulator cell; a glass tubecontaining 25 cells connected in series (laboratory design often used, inter alia, duringthe war in clandestine radio receivers); .(fourth row) three units with square cells connected in different ways.

load. These two fields will be discussed here briefly,confining our remarks, as far as the first field isconcerned, to voltmeters.The circuit of an alternating voltage meter

with rectification is represented in fig. lla. Themoving-coil instrument is connected to the D.e.terminals of4 rectifiers in Gr a et z circuit. The alter-nating voltage to be measured is connected to theseries connection of a resistor and the two otherterminals of the G r act z circuit.

of a rectifier (twice because the current alwaysflows through two rectifiers in series). Actually,however, the rectifiers do possess capacity. Thisfact can be reckoned with by taking into accountthe capacity C drawn in fig. lIa in a brokenline. The alternating current I now introducedin the Gr a e t z circuit differs from the current 10that it is desired to measure. From the re-placement scheme of fig. lIb it follows, after asmall calculation, that:

273

274 , PHILlPS TECHNICAL REVIEW , ,1947/1948

Let us first consider the case where the seriesresistance R is large compared with T (this will indeedbe so in the case of meters for not too low voltages;

'V V

0-

l2. 50953

Fig. H. ei) Circuit for measuring alternating voltages with theaid of a moving-coil -meter M, a Graetz circuit of four cells

, 'and a series, resist?.r R. By means of the broken line capacityC allowance can be made for the capacity of the cell. '

b) Replacement scheme where T represents the resistance ofthe meter and of two cells connected in series. V = the alter-nating voltage to he measured, I = the current, the' averagevalue of which is measured.

the' meter for 50 V, for instance, giving full deflec-tion at a current of 1 mA, already has R = approx.50000 Ohms, whilst r = approx. 1000 Ohms). If,moreover" WeT ~ 1, then for equation (1) one mayby approximation ,write:

_I R:i Jo ~1_:_HOJeT)2~. . . . . . (2)

With r again 1000 'Ohms 'and e= 1000 pF ,~e find,f~r a frequency of 50 cfs WeT R:i 3'10-4, so that thèfrequency error ~f2 (wer)2 is only about 5'10-8, thusamounting to 5'10-6%. For a frequency of 50 OOO,cfshowever, WeT R:i 0.3, corresponding to an errorin indication of about 5%. Therefore, to keep theerror below a certain- value within the largest pos-sible frequency range one must make er as smallas possible. Now e is-proportional to the effectivesurface ofthe rectifier, so that it is obvious to select,a small surface. This is, it is true, accompanied by aliigher value of T but the increase of T is much "lessthan proportionàte to the rednetion of the surface,in the first place because r = Tm + 2rd partly con-sists of the unchanged resistance Tm of the moving-coil meter and further because the rectifier resis-tance Td increases less rapidly than one wouldexpect. This last feature is related to the curvatureof the characteristic, which decreases according as

(1)'the.,'current density is raised. Fig. u illustrates thatwhen halving the effective 'surface of a rectifier' thevoltage loss, for the same current, increases by lessthan a factor 2.The relation between the momentary values of

voltage loss and current is not constant; Td is meantas an average value of this relation. It is thereforeclear that for given valves rd will depend upon thecurrent amplitude.This brings us to the deviations from the

linearity of the scale, which are particularlyevident in the case of voltmeters for low voltageswhere R ~ Tm + 2Td no longer applies, so that thenon-linearity of Td has some effect. Here, too; it istherefore of importance to select a high currentdensity and thus to use small rectifiers 5).This explains why the, diameter of the discs of

type I -. which have been particularly developedfor use in combination with moving-coil meters -has been fixed at only 1.2 mm.Before we leave the application of selenium rec-

tifiers in meters let us consider for a moment thesafeguarding of meters against overload.Let us suppose that wc have a rectifier connected

r

_V

50954

Fig. 12. (1) and (2) are characteristics of the current i,in thetransmitting direction as func.Ions of the voltage v of twosimilar cells whose effective surfaces are as 2 : 1. The voltageloss in the cells when the same current passes through (currenti represented on the left as function of the time t) is given bythe curves (3) and (4). The amplitude of (4) is less than twicethat of (3).

,5) From this it follows that a meter will show a smaller fre-quency error at the full deflection than at a smaller one.As regards the influence of frequency in the case of Iowvoltage meters -this is less according as R .is smaller com-pared with T, as is easily seen both from fig. 11 and fromequation (1).

\,

Vol. 9, No. 9 .'1 SMALL 'SELENIUM RECTIFIERS 275

in parallel to a moving-coil instrument in the man-ner shown infig. 13a. The diniensions of the meterand the rectifier are such that the voltage loss oc-'curring at full deflection lies below the threshold

a b 62 50955

Fig. 13. a) Direct current meter M safeguarded against over-load by the rectifier G connected in parallel. ,

b) For the proteetion of alternating current meters tworectifying cells (G1, G2) are used, connected anti-parallel.

voltage of the rectifier .. The resistance of the rec-.'tifier is then much higher than that of the meter,'s~ that the connection of the rectifier does not influ-ence the deflection of the meter.' Nów if th~ totalcurrent I (fig. 13a) increases to such an extent thatin the absence of the' rectifier the meter would be..damaged, then by connecting the rectifier in parallelthis danger is considerably reduced. The fact is thatwith sufficiently high voltage the resistance of therectifier drops to a fraction of the meter resistance,so that only a small portion of the total currentpasses through the' meter. A normal type of meterfor 0.1 mA for instance has a resistance of 15000Ohms, so that for the full deflection 0.15 V is re-quired. At th~t voltage the resistance of the recti-fier is approx. 0.5 Mohm, thus more than 300 timesas high as the meter resistance. With a current 6times as strong flowing through the meter - whi~hit can withstand for some time - the voltagebecomes 0.9 V, at which level, as follows from thecharacteristic of the rectifier, more than 9 mAflows through the rectifier and the total currentthus amounts to approx. 10 mA, i;e. 100 times thenominal value.

For th~ safeguarding of A.C. meters two rectifiersare used in anti-parallel connection '(fig. 13b).A somewhat analogous application of selenium

rectifiers has been previously described in thisjournal 6), where use has been made of the shape ofthe blocking characteristic to get a linear decibelscale on a measuring instrument.

2) Selenium rectifiers in modulators

In carrier-telephony circuit elements are needed

8) F. de Fremery and J. W. G. Wenke, The measure-ment of peak voltages in a studio installation, PhilipsTechn. Rev. 7, 20-23, 1942.

that have a non-linear characteristic, both onthe transmitting side in order to 'modulate thelow-frequency audio vibrations on one of the car~ierwaves, and at the receiving end for the' reverseprocess. As already described in this journal 7),selenium rectifiers lend themselves for this purpose,for instance in the so-called double push-pull circuit(fig. 14). 'It is to be recalled 'that on the output side of this

circuit certain 'undesired components (inter aliathose with the carrier-wave frequency itself), whichotherwise occur in other circuits are absent here,A condition is, however, that thé four rectifiersused in such a modulator must have the samecharacteristic and equal capacity. This condition isnetter satisfied with rectifiers of type I than thoseof type IJ originally used for this, purpose. More-over, type I has the advantage of a lower threshold

'voltage and a smaller differential resistance in thetransmitting direction, so that a smaller carrier-wave power suffices. In the third place type I con-stitutes an improvement because at voltages abovethe threshold value the characteristic is less curved;as a result tlie output voltage of the modulator isless dependent upon fluctuations in the carrier-waveamplitude ..Cellsof the type III also find application as modu-

lator cells, where higher voltages are required thanthe other types are able to withstand. Such is thecase with the modulator in the so-called signalreceiver; this is a component part of the signallingmechanism .in carrier-telephony 8).

3) Selenium cells in small rectifiers

We will not conclude this summ,ary without at

3

4

tp5. 6

JIOJJ

Fig. 14. Modulator in double push-pull circuit. Voltage withthe audio frequencies q is applied to the terminals 1 and 2,whilst voltage with the carrier frequency F is applied to theterminals 5 and 6..The modulated voltage is taken off at ter-minals 3 and 4. '

7) F. A. de Groot and P. J. den Haan, Modulators forcarrier telephony, Philips Techn. Rev. 7, 83-91, 1942.

8) In the case 'of a telephone link signalling is understood tomean the apparatus required for exciting and transmittingsignals for calling, dialing, etc; see F. A. de Groat,Signalling in' carrier telephony, Philips Techn. Rev. 8,168-176, 1946.

276 PHILIPS TECHNICAL REVIEW 1947/1948

least mentioning an important but only vaguelydefinable field of application of small selenium cells.We refer to those cases where the voltage obtainedby rectification serves, for instance, for the activa-

Fig. IS. Cascade connection consrstmg of foul' tubes, eachcontaining about 25 selenium cellsin seriesçand four capacitors.With this device the direct voltage of ] 200 volt is obtainedfrom' 220 V alternating voltage.

tion of a relay or as grid or anode voltage of anamplifying cell. With the supply and charging rec-tifiers discussed in one of the articles already quoted(seefootnote 2) it is of course impossible to indicatea sharply defined limit. The applications in questionare more or less incidental. Several of them havealready been mentioned at some place or other inthis journal; for instance there is the cascade circuitillustrated in jig. 15, which transforms 220 V alter-nating voltage into 1200 V direct voltage withouta transformer 9), and further the rectifier for sup-plying the anode voltage in radio receivers withextremely small dimensions 10).Where it is a matter of voltages of some tens of

volts or more, rectifiers with an artificial blockinglayer, type Ill, will as a rule be indicated, since theycan bear higher voltages in the blocking directionthan the other types, so that a smaller number ofthem are needed.

9) Philips Techn. Rev. 6, 78, 1941.10) Philips Techn. Rev. 8, 338, (fig. 3), 1946 (No. 11).