1959/60, No. 10 309

AUTOMATIC CONTROL OF A FILAMENT-COILING MACHINEWITH THE AID OF PRESET COUNTERS

by F. EINRAMHOF and P. HAVAS. 621.326.652.3:621. 778.3-52

In manufacturing certain. types oJ filaments for incarulesceni lam.ps the filameni-coilingmachine is usually corurolletl by a cam mech.anism: This method is not p articularly convenientwhen it is desired from time to time to switch the production to different types ofjilameni. Thisarticle describes an electronic control circuit incor poraiing cold-cathode tubes, in which thelengths of the coil sections can be adjusted by means of switches.

Special-purpose incandescent lamps, e.g. for stu-dio lighting, film projection and coastal navigationlights normally have filaments composed of a num-ber of coiled sections spaced by sections of uncoiledwire (fig. I).

99832

Fig. 1. Filament composed of four parallel coiled sections(See Th. J. J. A. Manders, Incandescent lamps for film pro-jection, Philips tech. Rev. 8, 72-81, 1946, page 74.)

There are two methods of manufacturing thesefilaments. In the first method the wire is coiled intoa continuous helix. Afterwards the straight sectionsare made by straightening one turn at the appropriatelocation. If the straight sections have to be longerthan the length of a single turn, the second methodis adopted. Here the straight parts are producedautomatically: the winding machine produces a coilinterrupted by straight sections of wire.

For the second method the winding machine mustbe controlled according to a certain programme,which was hitherto realized mechanically bymeans of cams. For changing over to the manufac-ture of another type of filament, certain gears in thecamshaft drive have to be exchanged. For large runsthis can hardly be called a drawback, but for small

runs a more flexible form of control is preferable. Itis all the more desirable for development work onfilaments for new applications or for improvementsto existing types. One important problem is attain-ing the optimum temperature distribution withinthe filament. For systematically investigating thismatter, the lengths of coil sections and gaps must bevaried independently. With cam-mo tion control thiswould become a costly and elaborate operation.

An electronic device using preset decade counters,i.e. counters giving a voltage pulse when reaching apreset number, has now been designed for thecontrol of filament-coiling machines. The preselectedprogramme can be easily and rapidly changed bymeans of a few switches. An experimental set-up ofa simple filament-coiling machine controlled by thiscircuit will be discussed here.

Description of the coiling machine

A sketch of the coiling machine is shown infig. 2.A bobbin S with tungsten wire is fitted to winding

o

99856

Fig. 2. Sketch of the winding machine. M motor. W windinghead. S bobbin with tungsten wire to be wound on mandrel D.F £lange fitted to the winding-head spindle. E electromagnet,which, when energized, pulls tbe flange against clutch plate K,so that the winding head is driven by the motor. V spring,pressing the flange to braking disc R when the electromagnetis not energized, thus stopping the winding head.

310 PHILIPS TECHNICAL REVIEW VOLUME 21'

head W.When electromagnet E is energized, flangeF is pulled to the right against clutch plate 1(; thewinding head is now driven by the motor. When Eis no longer energized, the disc is pressed by springV to braking discR and the winding head is stopped.The wire is coiledaround a mandrel D moved axially,through the winding-head spindle by the motor.(This mandrel is subsequently dissolved in acid, sothat only the coil remains.)Winding a filament composed of, e.g., three coiled

sections and three straight p~rts is done in thefollowing stages.a) The motor is running and the electromagnet is

energized, hence the mandrel is moving and the"winding head is rotating, so that the first coiledsection is being wound.

To stop winding, the current through the electro-magnet must be interrupted. If this were to happenwith the motor running at full speed, some slipwould occur because the hrake cannot immediatelybring the head to a standstill. The pitch of the coilwould therefore increase and the coil would be toolong. Besides, the slip of the head is not alwaysuniform, so that deviations from the nominallengths cannot be prevented by interrupting thecurrent at an earlier moment. The speed of thewinding head must therefore be reduced first. Thiscomprises the next stage, viz.:b) The motor is switched off, whilst the electro-

magnet remains energized. The motor is runningout hut winding is still continuing at decreasingspeed.

When the speed of the winding head has droppedsufficiently, the electromagnet is cut out, stoppingthe winding head. At the same time the engine isrevved up again for the third stage.c) The motor is running, but the electromagnet is

not energized. The mandrel is now moving butno winding takes place, so that a straight sectionof the filament is heing formed.

The coil is thus made in stages (a) and (b), thestraight part in (c). For the next coil-and-straightsection these three stages are repeated, so that afilament comprising three such sections is made innine stages. The velocities of motor spindle andwinding head· during four successive stages areshown in jig. 3.

Principle of the control circuit

Each stage must he switched on as soon as thepreceding stage is complete, i.e. as soon as the coilhas the necessary number of turns or the straightsection has the determined length. This implies, ineach case, that the motor spindle has then completed

nw

1

(oooooooooopooooo'II

rOOOOOOOOOO)I :I IIIIIIII

99867

:--tIIII

a c a

-tc Ga

Fig. 3. Rotation speeds nm of the motor spindle and nw of thewinding head as functions of time: a coil winding at full speed,b winding at decreasing speed, c formation of straight sectionof filament. The coil thus produced is shown schematicallyabove the diagrams.

a certain numher of revolutions. If the entire processis to take place automatically, a device is requiredto count the number of revolutions of the motorspindle, and to produce a switching pulse when thisnumber is reached. These numbers must be readilyreset to make it possible to vary the lengths of thecoils and straight parts independently of each other.

For counting the number of revolutions of themotor spindle the latter is fitted with a disc P withten apertures (jig. 4). On one side ofthe disc is a lampL, on the other side a phototransistor Tr whichtriggers a transistorized pulse generator I ten timesper revolution. The pulses from I (height 80 V,length 25 fLsec)are applied to a decade counter Te,consisting of four decade-stages of cold-cathodetubes. The maximum counting capacity is 9999pulses, or a 1000-turn coil, with an accuracy of1/10 turn.

The counter is connected to a programmer unitPB, likewise comprising cold-cathode tubes. Whenthe preset numher of pulses has heen counted, theprogrammer applies a signal to switching unit SBupon which the latter energizes motor and)orelectromagnet for the next stage. At the same timethe counter is reset to zero, so that the number ofpulses of the next stage can he counted.

p 99868

Fig. 4. Block diagram of the winding machine with electroniccontrol. W winding machine. M motor. P perforated disc onmotor spindle. L light source. Tr phototransistor. I 'pulsegenerator ..Tecounter. PB programmer unit. SB switching unit ..

1959/60, No. 10 AUTOMATIC CONTROL OF A FILAMENT-COILING MACHINE 311

The counter circuit described here thus gives aswitching command when the preset number isreached, unlike decade counter tubes such as theElT 1), which only indicate the number of pulsescounted.

The counter

The cold-cathode gas-discharge tubes employedare of the type Z 70 U (figs Sa and b). Apart fromcathode and anode, this tube contains two auxiliaryelectrodes, viz. a trigger electrode and primingelectrode arranged as second cathode. This latter

99799

a bFig. 5. a) Cold-cathode gas-discharge tube, type Z 70 U.b) Schematic diagram. A anode. K cathode. 5 trigger electrode.I priming electrode. A small dischmge current is permanentlypresent between priming electrode and anode, to minimize thedelay in establishing the main discharge.

electrode is fitted very close to the anode, so thata priming discharge is initiated by an extremelysmall current (3 [.lA);to limit the priming current tothis small value the priming electrode is earthed viaa high resistance. The priming discharge ensuresthat ions are always present within the tube, so thatthe main discharge is established more quickly. Thetrigger voltage (trigger-cathode) to initiate the maindischarge is 145 V; the burning voltage betweencathode and anode is 120 V.

Each decade stage comprises 10 tubes, circuitedas shown in fig. 6. The +80 V pulses to be countedare applied to input J. Consider the situation whenonly valve 0 is conducting. This tube then has acathode potentialof 100 V; the trigger electrode oftube 1 has the same potential. When the next

1) See the article: A. J. W. M. van Overbeek, J. L. H. Jonkerand K. Rodenhuis, A decade counter tube for high countingrates, Philips tech. Rev. 14, 313-326, 1952/53.

The preset decade counter employed and the program-mer-tube circuits described below were developed byL. Wasser and E. Strauman of Philips AG Zurich.

A similar counting circuit for an entirely different appli-cation was described some years ago: J. Domburg andW. Six, A cold-cathode gas-discharge tube as a switchingelement in automatic telephony, Philips tech. Rev. IS,265-280, 1953/54.

------------------

counting pulse reaches the input, the triggeringvoltage of tube 1 is exceeded, so that now this tubeignites. The pulse is likewise supplied to the triggersof the other tubes, but thesc do not ignite as theyhave as yet no trigger bias.When tube 1 ignites, tube 0 extinguishes. The

cathode of tube 1 initially remains at earth potentialowing to the presence of Cb, so that a large currentflows through tube 1. This current temporarily setsup a high voltage across the common anode resistorRa, so that the anode voltage of all tubes drops. Thecathode of tube 0, however, is still maintained at100 V by Cko, so that the anode-cathode voltageof this tube drops below the burning voltage. Whilsttube 0 extinguishes, tube 1 remains conductive,because the cathode potentialof the latter tube is asyet low, so that here the anode-cathode voltage doesnot fall below the burning voltage. While the voltageacross Ckl rises towards 100 V, the current throughtube 1 decreases.Because tube 1 is now conducting, a 100 V bias is

applied to the trigger of tube 2, so that upon the nextpulse the latter tube ignites and tube 1 extinguishes.Each successive tube is similarly ignited. The num-ber of the tube conducting at any given momentthus indicates thc total number of pulses received.

The maximum counting rate is determined by two factors.Once a tube has been ignited by a counting pulse, the nextcounting pulse must not arrive before the cathode voltage ofthis tube (and hence the hias of the next tube) has reached100 V. It is accordingly the RC-time of the cathode circuitthat determines the counting rate. This is moreover restrictedby the time necessary for initiating the discharge. The presentcircuit has a maximum counting rate of 2000 pulses/sec.

When tube 9 is conducting, the next pulse has toinitiate two events. First it has to ignite tube 0 ofits own decade. This means that tube 0 mustreceive its bias from tube 9. The ten counter tubesare accordingly arranged as a ring circuit. Secondlya pulse must be applied to the next decade. That iswhy the first decade is equipped with an additionaltube, the "relay" tube D (fig. 6), arranged in a self-quenching circuit. D is biased when tube 9 is con-ducting, and will be fired by the next counting pulse.The relay tube, however, is not directly connected tothe common anode resistor of the decade, but via asecond resistor Ra' of sufficiently high value toprevcnt a sustained discharge within the tube. Acapacitor Ca is connected in parallel with the anoderesistors, allowing the tube to fire for a moment, butquenching it as soon as the voltage across thiscapacitor has been built up. This means that thecathode voltage just reaches 100 V for a momentand then drops to zero. The latter pulse is trans-

.,.,---~------------~~-~~ ~~-

312 PHILIPS TECHNICAL REVIEW VOLUME 21

mitted from the output 0 to the input of the nextdecade.In this way the four decades are connected in

cascade. The indication of the counter is given by

+290V

The charging current of this capacitor causes avoltage pulse across resistor R that will fire the tubeif the trigger electrode has the necessary bias.

The operation of the complete circuit of counter

99670

Fig. 6. Decade of a decade counting circuit employing cold-cathode tubes (Z 70 U). 0-9counting tubes. D relay tube. I input. C input capacitors. Rn common anode resistor.Rn' additional anode resistor for tube D. Cn reservoir capacitor whereby D passes a shortcurrent pulse when fired. 0 output to the next decade. N input for the reset pulse, forreturning the counter to zero.

the trigger tubes themselves which light up whenignited, behind windows with the appropriatenumbers.

For resetting the counter to zero, a +300 V pulseis applied (at N) to tube 0 of all four decades. This issufficient to ignite these tubes even if they have nobias, whilst the temporary drop in anode voltageextinguishes all other tubes. Reset to zero is com-manded by the programmer unit, which produces a100 V voltage pulse for this purpose. This voltage isapplied to the zero-reset tube, a cold-cathode tubesupplied with a permanent bias from a voltagedivider. Like the relay tube of a decade, it isarranged in a self-quenching circuit and thereforeproduces only a pulse -of current, This current isstepped up to the required 300V by a transformer.

The programmer unit

The programmer unit also incorporates cold-cathode tubes of type Z 70 U, viz. one tube perdecade for each stage of the coiling process. Thecircuitry of such a programmer tube is shown infig. 7. The tube gets its trigger bias from anotherprogrammer tube. The trigger of the tube is connec-ted, via an input capacitor C and a Iû-positionswitch, to the cathode of one of the counting tubesof a decade. When the decade reaches the preselectednumber on the switch (except zero, see below), avoltage of +100 V relative to earth is applied to C.

and programmcr tubes may now be explained withthe aid ofthe block diagram oî fig. 8. Like the tubesof each counting decade, the programme tubesPI-Pas are arranged in a ring. The tubes PI-P4

correspond to the number indicating the desiredlength of the first coiled section of the filament; letus assume that in the counter the number 6527 has

--------"J_99871 --;;:

Fig. 7. Cold-cathode tube asprogrammer tube. The ten-positionswitch S connects the trigger electrode of the tube to -thecathode of one of the counting tubes of the decade. The tubeis biased via the cathode of the preceding programmer tubeand then fires when the counter indicates the appropriatenumber. The programmer tubes have a commonanode resistorRn.

1959/60, No. 10 AUTOMATIC CONTROL OF A FILAMENT-COILING MACHINE

been preselected. This means that the trigger of PIis connected to the cathode of tube 6 of the thou-sands decade, P2 to tube 5 of the hundreds decade,P3 to tube 2 of the tens decade and P4 to tube 7 ofthe units decade. Before the winding process isstarted, the counter is set to zero and the bias is

which changes the activation of motor andmagnet as required for ending the first stage inthe winding process and starting the secondstage.

2) The voltage is applied to the zero reset. Thecounting decades are thus set to zero and are

Fig. 8. Block diagram of the control circuit. Dl' DlO' DlOO' DlOoo decades of the counter.1 pulse generator. N zero resetter. St starting circuit. PI-P36 programmer tubes. 51-59switching tubes. TR Schrnitt trigger circuits. K power transistors. JWmotor. E electro-magnetic clutch. Black connections are those for the count pulses, firing voltages andswitching voltages; blue connections are for the bias voltages; red connections are for thezero reset voltage; green connections are for the starting pulse.

applied to PI' The motor is now started, so thatcount.pulses are applied to the input of the counter.When the thousands decade changes to 6, PI ignites.This produces the bias for P2• Now the next occasionthat the hundreds decade reaches 5, P2 can fire.6500 pulses have now been counted. PI extinguishesagain and P3 is biased. Then, after 6520 pulses, P

3will fire and similarly, after 6527 pulses, P4•

The cathode of P4 then has a voltage of 100 V.This is used to start the next stage as follows.1) The voltage is transmitted to a switching unit,

immediately ready for counting the secondnumber of pulses.

3) P5 is biased.Whilst the second stage is being completed, valves

P5-P8 are successively ignited, after which the thirdstage is started. In this way the entire programmeis completed. The last tube of the ninth stage appliesagain the bias to PI' so that production of the nextfilament starts without interrup tion.In position 0 the ten-position switches have afunction different from that described above. Con-

313

314 PHILIPS TECHNICAL REVIEW VOLUME 21

sider the case in which 6501 pulses have to becounted for the first stage. PI will then normally fireat 6000, P2 at 6500. For 6501, P3 should fire simul-taneously with P2 and supply the bias to P4, sincethat tube has to fire at the next count pulse. If thetrigger of P3 (with the corresponding switch set toposition 0) were connected to the cathode of tube 0of the tens decade, however (as corresporids to otherpositions of the switch), P3 would not fire at all.This is because after 6499 pulses it is first the unitdecade of the counter that changes to zero and thenthe tens decade, and only then does the latter supplya pulse to the hundreds decade, changing it to 5.This fires P2 and only then will P3 be biased. P3,

however, can no longer fire, because tube 0 of thetens decade is already conducting and the ignitionpulse is no longer available. In this case it is accor-dingly necessary to bypass P3 in the circuit; P4 mustobtain its bias directly from P2• It is thereforearranged that when a ten-position switch is inposition 0, the bias received from the precedingprogrammer tube is directly relayed to the next

programmer tube.

The switching unit

Next to the programme tubes in fig. 8 are shownthe switching tubes SI-S9' one for each stage of thewinding process. These nine tubes are similarlyarranged in ring configuration. The cathode of cachtube is connccted to either one or two Schmitttrigger circuits, which are triggered whenever thecorresponding programmer tube ignites, therebycontrolling the current to the winding motor and/or

magnetic clutch via power transistors.To start the whole winding process, a 300 V

pulse is required. This pulse is obtained by dis-charging a capacitor across a resistor when a push-button is depressed. The pulse ignites SI and Pse-The two Schmitt circuits connected to SI are thentriggered, energizing motor and magnet for the firststage (coiling). The cathode voltage of P36 hasthree effects: 1) zero-re setter N is activated; 2) PI isbiased and the first number is now counted off asdescribed above for the programmer unit; 3) thevoltage is applied to the trigger electrode of SI(which is necessary at the end of the programme).Because SI is now conducting, S2 is biased. Thecathode of P4 is connected to the trigger electrodeof S2. When the first number is reached, P4 fires andS2 fires immediately afterwards. This extinguishesSI' and motor and magnet are cut out. The Schmittcircuit following S2' however, is immediately trigge-red, thus directly energizing the magnet again. Themagnetic clutch does not respond to this very short

interruption of the current. The motor remainsswitched off, and winding is slowing down. P4 hasmeanwhile also activated the zero-re setter andbiased P5' so that now the second number is countedoff. Upon completion, Ps ignites, so that S3 (biasedby S2) fires: the third stage therefore commences(motor only switched on, forming straight section offilament). During the last stage, SI is biased and cantherefore be ignited by P36' enabling the programmeto be repeated without interruption, i.e. to wind asecond filament, and so on. During each stage thecorresponding switching tube remains conducting,so that there is always visual indication of the stage

in progress at any moment.By means of the 36 ten-position switches the

9 numbers can be independently varied, which

makes a highly flexible system.

Construction of the equipment

The major part of the circuit is accommodated inboxes of 40 X 104,X 117 mm. Printed wiring is usedfor all component parts, and connections betweencontainers are established by contact strips formingpart of the printed wiring. The contact strips fit intospecial sockets arranged on a panel. Each decade ishoused in its own container, togethcr with twoprogrammer tubes anel the two corresponding digitswitches (fig. 9). The position of the counter isindicated through thc windows at the front. Thc

.99800

Fig. 9. Two views of a decade of the counting circuit; eachdecade chassis also houses two programmer tubes and theirten-position switches. If more than two stages are to be pro-grammed (nine in the case described here), additional pro-grammer tubes are available (for each decade). The numberedwindows behind which tbe counting tubes are fitted can beseen on the front of the decade boxes. The tubes are paintedblack to prevent premature firing by incident light. The decadeswitches together with figures indicating their positions canalso be distinguished.

1959/60, No. 10 AUTOMATIC CONTROL OF A FILAMENT-COILING MACHINE 315

digit switches can be manipulated from the frontand their positions read. The remaining programmertubes are accommodated in pairs or groups of four infurther boxes, together with their switches, and maybe independently connected up to a decade. Theswitching tubes are housed in a separate box. Theauxiliary circuits, being likewise made with printedwiring, can be similarly incorporated into the circuit.

The equipment described here has possibilitiesbeyond that of controlling a coil-winding machine.Generally speaking, it can be employed wheneversome switching action has to follow a specifiednumber of occurrences, as long as these occurrencescan be translated into light or voltage pulses. Asexamples we mention the control of machine toolsby counting the number of revolutions of driveshafts, and the counting and sorting of objects onbelt conveyors.

Switching can also take place after specified timeintervals. The circuit is then fed with pulses from apulse generator with a known and constant frequen-cy. This possibility has been used for coating bulbswith an internal mirror. This is done by vaporizinga piece of aluminium within the evacuated bulb,_after which the aluminium vapour settles as adeposit on the wall. The times required for evacua-

ting the bulb, for heating the aluminium and forcooling down, depends upon the type of bulbtreated. With the apparatus described these timescan' be rapidly preselected a~d changed. For smallruns, especially, the flexibility of the apparatusagain shows up to advantage over mechanicalcontrol.

Summary. In various incandescent lamps, e.g. those forprojectors, the filaments consist of coiled sections linked bysections of uncoiléd wire. Filaments of this kind can be madeon a filamentwinding machine by decouplingthe winding headfrom the motor drive for a few specifiedperiods. Motor andcoupling must be energized in accordance with a preselectedprogramme, according to the type of filament required. The'machine was hitherto mechanically controlled by cams. Anelectronic control has now been designedfor the same purpose.It has the advantage that for small runs, e.g, in developingnew types, the dimensions of the filament produced can easilybe varied. The control is operated by voltage. pulses derivedfrom the periodic interruption of a light beam by a perforateddisc fitted to the motor spindle. The pulses are applied to adecade counter circuit incorporating cold cathode tubes, typeZ 70U, as switching elements. Upon reaching a preset number(set by a switch in each decade), corresponding to a givenlength of coil (number ofrevolutions ofthe motor), the countercontrols a transistorized switching unit, which switches thecurrents through motor and magnetic coupling. In this waythe various stages of the winding programme are successivelycompleted. The control apparatus is in principle also applicableto other manufacturing processes, e.g. in which the preselectednumbers correspond to time intervals. As an example aninstallation for the internal silvering of bulbs is mentioned.

ABSTRACTS OF RECENT SCIENTIFIC PUBLICATIONS BY THE STAFF OFN.V. PHILIPS' GLOEILAMPENFABRIEKEN

Reprints of these papers not marked with an asterisk .. can be obtained free of chargeupon application to the Philips Research Laboratories, Eindhoven, Netherlands.

2698: P. B. Braun and J. L. Meijering: The copper-rich part of the copper-barium system (Rec.Trav. chim. Pays-Bas 78, 71-74, 1959, No. I).

The system Cu-Ba was examined up to 75% Baby weight. At 675°C there is a periteetic three-phaseequilibrium Cu + liquid ~ Cu13Ba. The compoundCu13Ba has KZn13 structure. At 550°C there is aeutectic three-phase equilibrium liquid ~ Cu13Ba+ CUxBay-Composition and structure of the lattercould not be determined owing to experimentaldifficulties.

2699: W. Kwestroo: Spinel phase in the systemMgO-Fe203-Al203 (J. inorg. nucI. Chem. 9,65-70, 1959, No. I).

The three-component system MgO-Fe203-AI203

has been investigated at 1250 °C and at 14.00°C. The

binary parts of this system, already described sometwenty years ago, are reviewed and completed. Theternary diagram shows a broad spinel area: this areaincreases with increasing temperatures. The pre-parations do not contain Fe2+ ions, as was ascer-tained by analytical methods andbyD.C.resistivitymeasurements. The resistivity has a fairly highvalue. The physical properties are roughly in agree-ment with the properties already found in systemsinvestigated previously. Substituting AI3+ ions forFe3+ions in the spinel phase lowers the value of themagnetic saturation and the Curie temperature. Anincreasing amount of Fe203 in the spinel phaseincreases the magnetic saturation and the Curietemperature. Samples with the composition of amineral called "hoegbomite" have been investigatedand the possible structure of this mineral is discussed.

316 PHILlPS TECHNICAL REVIEW VOLUME 21

2700: J. Goorissen and F. Karstensen: Das Ziehenvon Germanium-Einkristallen aus dem"schwimmenden Tiegel" (Z. Metallk. 50,46-50, 1959, No. I). (Pulling germaniumsingle crystals from a floating crucible; inGerman.)

Single crystals of germanium with a homogeneousimpurity concentration can be prepared if enrich-ment or exhaustion of the impurity in the melt isprevented. This is made possible by means of thefloating-crucible technique. In this technique a smallcrucible from which the crystal is pulled floats onthe molten germanium in a larger outer crucible.Communication between both crucibles is madepossible by means of a capillary allowing a contin-uous replacement of the solidified germanium. Thetheoretical yield of this technique is compared withthat of the Czochralski method and of zone levelling.See also Philips tech. Rev. 21, 193-195, 1959'60(No. 7).

2701: H. J. G.Meyer: Theory ofinfrared absorptionby conduction electrons in germanium (Phys.Chem. Solids 8, 264-269, 1959).

It is shown that a theory of infrared absorptionby conduction electrons which takes into accountthe structure of the conduction band and acousticalas well as optical intra-valley scattering can bedeveloped without any serious approximation. Incombination with an estimate of the possible influ-ence of impurity scattering the theoretical results canbe used for the determination of one of the acousticaland of the optical deformation potential constants.From available experimental data the approximatenumerical value of the latter is determined. Thegenerallimits of validity of the theory are discussed.

2702: K. W. van Gelder: Fabricagebeheersing,I. Procesnauwkeurigheid en het stellen vantoleranties (Sigma 5, 15-19, 1959, No. I).(Process control, I. 'Process accuracy andthe specification of tolerances; in Dutch.)

First of two articles (see also 2719) in which theauthor attempts to fuse the essentials of manu-facturing processes (derived from analysis of such

processes) with statistical concepts, and thus arriveat process control. This first article contains firstlya simple example of a process analysis (pressing andcentreless grinding of plastic coil bobbins), whichhe uses to draw conclusions as regards the choice oftolerances. This is further considered in a secondexample. Finally, more complicated cases are dealtwith in which so-called combined tolerances occur.

2703: G. H. Jonker: Analysis of the semiconductingproperties of cobalt ferrite (Phys. Chem.Solids 9, 165-175, 1959, No. 2).

From measurements of resistivity, activationenergy, and Seebeck effect, an energy-level schemeis derived by which the semiconducting propertiesof CoFe204 can he described. These properties differconsiderably from those of normal semiconductors,as the charge carriers are not free to move throughthe crystal lattice but jump from ion to ion.

2704: M. J. Sparnaay: On the additivity of London-Van der Waals forces (Physica 25, 217-231,1959, No. 3).

Calculations are given concerning the additivityof London -Van der Waals forces between two groupsof atoms, the atoms being represented as isotropicharmonic oscillators. The results indicate that de-viations of 10-30% from additivity can be obtainedif only dipole-dipole interaction between oscillatorsof one group is assumed. The effect can be expectedto be relatively large if the symmetry of the arrange-ments of the oscillators in the group is low, and it isdependent upon the relative spatial position of thegroups.

2705: G. Diemer and P. Zalm: The role of ex-haustion barriers in electroluminescent pow-ders (Physica 25, 232, 1959, No. 3).

Note concerning the apparent paradoxical effectsof Mott-Schottky barriers in phosphor grains in aninsulating medium. The explanation of the appar.entparadox lies in the fact that electroluminescenceoccurs only in tiny localized spots where exhaustionbarriers can indeed give rise to a local enhancementof the electric field.