304. PHILIPS TECllNJCAL REVIEW VOL. 14., No. 10·

AN OSCILLATOR WITU CONSTANT OUTPUT VOLTAGE

by L.,ENSING and H. J. J. van EYNDHOVEN. 621.396.615 :621.316. 7;22.1

Using the gelleral principle of a regulating system, It valve oscillator call be made to ~lIorkin suçh a way that the output voltage remains substantially constant. An application regardingan instaliasion. for the calibration of valve voltmeters is described.

For the generation of alternating voltages withfixed ore ontrollahle frequency, use is often made

I ,

of oscillators based on an electronic valve (triodeor pentode), an oscillatory circuit (Le circuit) anda feedback element. An oscillation generated in thisway can, however, show fluctuations in amplitudeIand in frequency and deviations from the sinusoidalwaveform which are most undesirable, especiallyfor measuring and calibrating purposes.

We shall describe below a method which is pri-marily directed towards making the amplitude ofthe oscillator independent of disturbing influencessuch as changes in the supply voltages of the valve,in the mutual conductance of the valve and in theimpedance of the Le circuit. With this method thevalve is working under conditions which are alsofavourable for the frequency to remain constantand for the generation of a purely sinusoidal voltage.

Regulating principle applied to an oscillator

A method on which many regulating systems arebased 1) consists in the comparison of the regulatedquantity x with a similar but constant referencequantity aref, in that the difference llre£-X is madeto control the apparatus in such a way thatlaref - xl is reduced.If the quantity which has to be kept constant and

the reference quantity are not similar, then anauxiliary quantity will have to be derived from thefirst mentioned one which is similar to aref. This isthe case in the matter under discussion: the alter-nating output voltage of a valve oscillator has tobe kept constant in amplitude (Vo)' but as referencevoltage we can of COUI'seonly take a direct voltage(Eref)' By rectification a direct voltage Eo which isproportional to Vocan be derived from the oscillatorvoltage and the difference Eref - Eo can functionas ;control voltage, i.e, it can control the oscilla-tor in such a way that the difference IEref - Eol,

. I',irrespective of its cause, becomes smaller. To attainthis, the control voltage can work, for example,as a bias in the control-grid circuitwith such polarity

1) See e.g, J.H. Ro osd o r p; On the regulation of industrialprocesses, Philips tech. Rev. Vol. 12, 221-227, 1951.

that the working point on the valve characteristicis shifted tI? a region of greater Ol' smaller' rrt.utualconductance accordingly as the voltage amplitudeVo was either too 'small Ol' too large"

Inflnence of the valve adjustment

Before elahorating on this thought it should herecognized that the limitation of the oscillationamplitude is effected by the working of the regula-ting system and not - as in the case of the usualLe oscillators with a grid capacitor and grid leak- by the occurrence of grid current. The latterbiases the control grid to such an extent that thevalve is cut off during part of the cycle (class Coperation) and the effective mutual conductancedecreases. The anode current thus becomes pulse-shaped and consequently contains strong harmonics;these harmonics also occur in the circuit voltage,although to a lesser degree.

By using a regulating system, however, one canensure that the valve remains operating in class A;thus operation is limited to a small, substantiallystraight part of the ia-vg characteristic, resultingin much less distortion of the circuit voltage.

The last is not only important if a sinusoidal out-put voltage is desired, but also because distorti?n isone of the causes of frequency fluctuations. AsHorton has shown 2), distortion results in a phaseshift in the voltage which is fed back to the grid,so that the frequency J for which the oscillatorycondition is complied with, is no longer the resonantfrequency Jo of the Le circuit. The differencebetween J and Jo increases as the distortion becomeslarger and will therefore not be constant if thesupply voltage fluctuates, if the valve heats up, andif it ages in the long run.

Another cause of the frequency shift may lie inthe well-known fact that the inter-electrode capaci-tances of a valve - especially the grid-to-cathodecapacitance - depend on the space charge. In thecircuit to be discussed the regulating system willkeep the direct current which flows through the

2) J. W. Hor t on, Vacuum tube oscillators. A graphicalmethod of analysis, Bell. Syst. tech. J.3, 509-524, 1924.

APRIL 1953 OSCILLATORWITH CONSTANTOUTPUT VOLTAGE 305

valve approximately constant in case of mainsvoltage fluctuations and when the valve ages, sothat the influence of this source of frequency fluctua-tions is reduced.

More detailed description of the regulating circuit

In various ways it is possible to effect the limita-tion of the amplitude, so that the oscillator remainsoperating in class A. In one well-known methoda small incandescent lamp 3) (or another devicehaving a resistance dependent on temperature) isused in one arm of a Wheatstone bridge; thisbridge can consequently only be in equilibrium atone particular value of the input voltage.

Another method has been stated by AI'guimb au 4),who uses a separate rectifier which' converts analternating voltage proportional to the oscillatorvoltage into a direct voltage, the latter serving asbias in the control grid circuit.

Taking the discussion of the regulating principlegiven above as a starting point, we can describeour method 5) as one in which two currents arecompared instead of two voltages. Like Arguim-

_ . b a u we use an auxiliary rectifier; it consists of adiode Dl and a blocking capacitor Ca (fig. la),

Vd

t --.t:V~~~--~--~----~

13178

Fig. La) Circuit for automatically stabilising the outputvoltage of an oscillator.P pentode. LC oscillatorycircuit.L' feedbackcoil.Ca blockingcapacitorandDl diode,rectifyingthe alternatinganodevoltageto apulsating voltageVd (repre-sentedin b as a functionof the time t). Erc£ referencevoltage.C8 grid capacitor.

3) L. A. Meacham, The bridge-stabilizedoscillator,Proc.Inst. Rad. Engrs. 26, 1278-1294, 1938.

4) L,B. Arguimb au, Ahoscillatorhaving a linearoperatingcharacteristic, Proc. Inst. Rad. Engrs. 21, 14-28, 1933.

6) This correspondsto a largeextentwith the triplex diodecircui t used in someradioreceiversfor delayedautomaticgain control; see e.g. "J. Deketh, Fundamentals of radio-valve technique I, pp. 329-332 (PhilipsTechnicalLibrary,1949).

connected in series between the cathode and theanode of the oscillating pentode P. Consequently apulsating voltage Vd is produced across the diode,with the anode negative with respect to the cathode(fig. 1b); the amplitude of Vd is 2Vo and its averagevalue Eo = - VOo A capacitor Cg is connectedbetween the points G and K (fig. la) of the gridcircuit; G is connected to the anode of the diodevia a resistor RI and to a point having a positivevoltage; the reference voltage Eref, via a resistor R2•

Currents 11 and 12flow through these resistors andwith a negligible grid current the. following formulaapplies:

11+ 12= O. . . . . _ . (1)

If Eg! is the direct voltage across the capacitor Cg,then 11= (Eo - EgI)/RI = -(Vo +EgI)/RI and12 = (Eref - EgI)/R2• Substitution of these valuesin (1) gives:

Rl RI+R2 -Vo = - Eref - --- Egi. (2)'

R2 R2

Egl - the control grid bias required to make thevalve oscillate with the amplitude Vo, at givensupply voltages and circuit impedance - dependson the valve characteristics. It is, however,possible to make the first term in the right-handmember of (2), in which Egl is absent, preponderateshongly with regard to the second term, for exampleby choosing Eref high with respect to EgI; in firstapproximation we then find that

RlVo=-Eref,R2

viz. III first approximation the amplitude of the.output voltage depends solely on the ratio of twofixed resistances and on one direct voltage, and con-sequently not on .the valve characteristics and thecircuit impedance. For a more accurate analysis,however, it is necessary to take into account alsothe influence of the last-mentioned quantities.

The stabilisation factor

The direct anode voltage, the screen-grid voltageand the heater voltage applied to the pentode,..and also the reference voltage .Eref, are as a ruleobtained from the A.C. mains and therefore showthe same relative fluctuations as the mains voltage,unless special measures have been taken to preventthis.It is obvious that, if Vo is to be kept' constant,

Eref should in any case be carefully stabilised, forexample by means of one or more voltage referencetubes. Fig. la shows that the screen-grid voltage

306 PHILIPS TECHNICAL REVIEW VOL. H, No. 10

can be obtained from the same voltage source with-out any difficulty; it is then no longer necessary tohe concerned regarding the influence of fluctuationson the screen-grid voltage.

The exact value of the direct voltage in the anodecircuit is not a factor of decisive influence; theproperties of a pentode are sufficiently independentof this voltage. Heater voltage fluctuatiöns, how-ever, result in variations in the mutual conductanceof the valve characteristic; they can be consideredtogether wi~h changes in mutual conductance dueto other causes. •

Finally, we have yet to consider the influence ofchanges in. the impedancè in the anode circuit.If the mutual conductance S changes. by an

, amount LIS (due to a change in theheater currentor by ageing of the valve), the original value isrestored by changing Egl by an amount

àEglLlEgl = -as LIS,

in which àEgiàS can be determined from the'S= f(EgI) characteristic. From equation (2) follows:

s, + R2 u,+ R2 àEglLIVo = - LlEgl = -à- LIS.

R2 R2 S

We now find for the ratio of the relative changeofthe mutual conductance to that of the amplitude,i.e. the stabilisation factor for mutual con-,ductance changes:

LlS/S Vo R2 esLIVel V« = SRI + R2 àEgl .

The alternating voltage (amplitude Vo)across theimpedance Z in the anode circuit is produced by thealternating anode current (amplitude la = Vo/Z),which in its turn is caused by an alternating voltage(amplitude VgI) on the control grid: la = SVgI•

Itwill be assumed that this alternating grid voltage,which is due to the feedback, is a fraction t of thealternating anode voltage: Vgl = t VOo To maintainthe oscillatory state, the known oscillatory conditionshould be complied with:

as=

SZt=-l,

which expresses that the total amplification for oneround trip through the oscillator should be exactlyequal to unity. •

The impedance Z is variable in oscillators withadjustable frequency; moreover it varies withchanges in the load on the oscillator and it canalso vary due to incidental causes, for example byincrease of the losses caused by dust and moisture.We see from (4) that changes in Z with constant t

are compensated by equal relative changes III S.Consequently the stabilisation factor for impe-dance variations,

az=jJZ/Z

LlV-j'V, ,o 0

is equal to the value of as found in (3).

For the sake of completeness we shall finally consider the_case of a n onvs t ab ilised screen-grid voltage.

It follows from (2) that a change LlEglof the control-grid •bias results in a change of the amplitude Voby an amountLIVo of the following magnitude:

Rl + R2. AER

2LJ SI'

For a pentode it holds that with constant anode current achange LlEg2 in the screen-grid voltage is compensated by achange LlEsl= - LlEg2/ftg2s1 in the control-grid voltage;ftS2Slis a factor larger than unity (e.g. 18 with a type EBF 80pentode). .A change in the screen-grid voltage might be thought to

[nfluence the mutual conductance. However, this is barely'so with a pentode; the characteristics of the anode currentas a function of Egl, with Eg2 as parameter, run almost paralleland consequently the mutual conductance which is requiredfor oscillating is reached at a certain value ofthe anode current-which is substantially independent of Eg2•

Consequently LlEg2/L1 Vo (the regulating factor) becomes

LlEs2 R2

LI Vo = Rl + Ro ftg2S1'

For the ratio of the relative changes, the stabilisa tionfactor for changes in screen-grid vol t nge, we find:

(3)

(4)

Numerical example

In agreement with a circuit we will discuss pre-sently, we have selected as a valve the EBF 80pentode ..Let Vohe (say) 40 V and Z= 30 kO, thenla is 1,33 mA. With t = 1/20, Vgl has to be 2 Vand the required mutual conductance is then S =1·33/2 mA/V = 0.67 mA/V.Keeping the screen-grid voltage constant 'at a

value of 85 V, the mutual conductance of 0.67 mA/Vis obtained, according to the valve data, at a gridbias Egl = -4 V, and amounts to àS/àEgl = 0.6mA/V2 at this adjustment.

Choosing also 85 V for the reference voltage, andRI= 1 MO, then from (2) a value of 2.5 MO followsfor R2•

According to (3) the stahilisation factor formutual conductance and impedance changes ·nowbecomes:

40 2.5 -3lasl = lazl = 3' . 0.6x 10 = 26.0.67 X 10- 1+ 2.5

APRIL 1<)53 OSCILLATOR WITH CONSTANT OUTPUT VOLTAGE 307

In this example the relative variations in the alter-nating voltage amplitude are consequently 1/26 ofthe relative mutual conductance or impedance varia-tions. This important improvement is obtained bymeans of only a few extra circuit elements. Consider-ably larger values of the stabilisation factor can beobtained by connecting an amplifier between theoscillatory circuit and the regulating device, aswill be shown presently.It should be observed that although stahilisation

may be obtained to a certain degree with a conven-tional LC oscillator by stabilising the supply volt-ages (including the heater voltage), changes inmutual conductance by the ageing of the valve andimpedance variations in the circuit may still occur.With the regulating circuit described above theseinfluences are also kept in check.

Moreover, the regulating system keeps the valveconstantly operating in class A. This not only offersthe advantage that the distortion of the circuitvoltage is very small - an advantage already men-tioned above - but also that the int erelectrodecapacitances of the valve remain constant whenthe mains voltage fluctuates or the tube ages, sincethe anode current is kept constant in that case.

Adjustment and modulation of the alternatingvoltage amplitude

By making the reference voltage Eref controllable,a simple means is obtained of varying Vo or adjus-ting it to a given value. The graph showing Vo asa function of Eref in accordance with equation (2)is gIven in jig. 2.

rovI6j

-E"ef 13179

Fig. 2. Graphic representatio n of equation (2): amplitude Voof the alternating voltage versus reference voltage Eccf forRI = 1 MO, R2 = 2.5 MO and Egl = - 4 V.

If Eref is varied around a certain value, theamplitude of the output voltage is modulated. Usecan be made of this if a modulated output voltageis required. Fig. 2 also represents the modulationcharacteristic, provided no grid current is flowing.If grid current does flow, equation (1) is no longer

applicable; the modulation characteristic thenbecomes less steep and consequently shows a sharpbend (jig. 3). Therefore the fact whether grid currentflows or not, detennines the greatest modulationdepth at which no distortion occurs.

Fig. 3. Oscillogram of the alternating voltage of an oscillatorin accordance with fig. la., which is being modulated with asinusoidal voltage. The voltage for the horizontal deflectionis likewise sinusoidal and in phase with the modulating volt-age. For registering this oscillogram, the valve was so adjustedthat grid current occurred, which caused a sharp bend inthe en velo pc.

Plotting the modulation characteristic (or displaying theoutput voltage on an oscilloscope) is the simplest mcthodof knowing whether grid current flows or not. The more usualmethod, i.e. to connect a voltmeter parallel to the grid resistor,would increase tbe stray capacitances to such an extent thatthe working point of the valve would be shifted considerablyin order to remain in compliance with the oscillatory condition,so that the measurement would then be taken under verydifferent conditions.

An upper limit is also set to the modulationfre quency viz. by the tendency towards squegging.

/(J]V

Squegging and starting

The eperation of the oscillator is determinedmainly by the magnitudes of the cap acitances Cg andCa (fig. la). The capacitor Cg in the grid circuit,for example, together with the capicitance betweencontrol grid and earth, forms a voltage divider; ifCg is too small, a large proportion of the voltageinduced in the feedback coil is lost in Cg. On theother hand, Cg should not be too large either - andthis also applies to Ca - since the control voltageis then built up so slowly that there is a risk ofsquegging; the normal oscillating process is thenperiodically interrupted at a frequency f~e], depen-ding on a relaxation time.It is possible to investigate whether there is a

risk of squegging by modulating the oscillator inthe above-mentioned way with an alternatingvoltage of constant amplitude and variable fre-

308 PHILlPS TECHNICAL REVIE\,r VOL. 14, No. 10

lfuency and by displaying the modulated outputvoltage on an oscilloscope. If the modulation fre-quency approaches .h·cl, a deepening of the modula-tion is observed. This puts an upper limit to themodulation frequency. A square-wave modulatingvoltage can also be used; the tendency towardssquegging can in that case be judged from thedamping of the oscillations present in the envelopeof the modulated output voltage wave (fig. 4).

Fig. 4. Oscillogram of the voltage across the LC circuit of" anoscillator in accordance with fig. la., modulated with a square-wave voltage. The tendency towards squegging can be judgedfrom the d a mping of the oscif la tio ns present "in the envelope.

If Cg anel Ca have bcen so chosen that the oscilla-tor (connected in accordunce with fig. la) can

+Eg2, fref

the resistor Rl and the diode Dl' while the greaterpart flows through the feedback coil and furtheras grid current through the pentode. Due to thelast fact the path between the control grid andcathode has a differential resistance which is muchsmaller than the impedance of Cg; consequentlyonly a small part of the growing alternating voltagewhich is induced in the feedback coil reaches thecontrol grid; moreover it does not have the correctphase. For this reason the oscillating process cannotstart.

To remove this obstacle, provision should bemade that I2 cannot flow as grid current. To attainthis, a diode D2 can be incorporated (fig. Sa),offering an easy path to I2' and a resistor R3' in-creasing the resistance in the path via the controlgriel.

Fig. Sb shows that a single valve, containing apentode system and two diode systems, such asthe above-mentioned EBF 80, can be used toadvantage. The reference voltage and the screen-grid voltage are stahilised at 85 V by one 85 Alvoltage reference tube 6). Consideration is beinggiven to the application of a system similar to thatjust mentioned in various measuring apparatusbrought on the market by Philips.

Fig. 5. a.) As fig. I a, but with the addition of a diode D2 and a resistor R3' which preventthe current '2 from Rowing across the control grid immediately after switching on; thiswould hamper the oscillator Irorn starting properly. Co smoothing capacitor. Meaningof the other letter references as in fig. Ill. b) Practical example of the principle with anEBF 80 valve, in which the pentode Panel the diodes Dl and D2 are combined. The ref-erence voltage, also screen-grid voltage, is kept constant by a 85Al voltage referencetube, with the series resistor R4'

oscillate without tendency towards squegging, itapparently has some difficulty in starting.The cause of this is the following: Immediately

after switching on, the valve by no means oscillatesat full strength and the alternating anode currentis still negligible; the current I2' due to the referencevoltage source, now flows for a small part through

Installation for the calibration of valve voltmeters

The stabilisation of an oscillator voltage has al-ready found practical application in an installationfor the calibration of valve voltmeters.

Fig. 6 shows this installation for the calibration

6) T. .lu rri a a n s e, A voltage stahilizing tube for veryconstant voltage, Philips tech. Rev. 8, 272-277, 1946.

APRIL 1953 OSCILLATOR WITH CONSTANT OUTPUT VOLTAGE 309

Fig. 6. Calibrating rack for the electronic millivoltmetertype GM 6016. 1 panel with four stabilised supply rectifiers.2 panel for the adjustmcnt of the supply voltage (50 cis) forthe meter to be calibrated. 3 panel with calibration voltagesfrom 3 mV to 300 V, 100 kc/s, for the calibration of meterGM 6016 and the accessory attenuator (Z2). 4 panel forchecking the frequency characteristic. 5 panel with twelveoscillators (frequencies 1 kc/s to 30 Mc/s), with automaticallystabilised output voltage. To the right, the drilJing machinefor drilling the catch holes in the calibrated attenuator.

of millivoltmeters type GM 6016, an instrumentsuitable for frequencies ranging from 1000 cis to30 Mc/s. The input voltage required to fully deflectthe meter amounts to 3 mV at maximum sensi-tivity 7); this voltage can be increased in stepsup to 1000 V by means of a capacitive at tenuator.A similar installation is in use for the calibration

of millivolt meter type GM 6005; this is a metersuitable for frequencies from 20 cis to 1 Mc/s whichgives full deflection at voltages from IOm V to300 V. In the following we shall only discuss thecalibration installation for type GM 6016.

7) It is only with regard to sensitivity that type GM 6016differs from type GM 6006 described before, which alreadygives full deflection at 1 mV (H. J. Lindenhovius,G. Arbelet and J. C. van der Breggen, Philips tech.Rev. 11, 206-214 1950.

Two checks can be carried out with the installa-tion under discussion:1) absolute calibration of thc meter including the

attenuator, and2) checking of the frequency characteristic.

An oscillator 00 for 100 kc/s (fig. 7) is mountedin panel 3 (fig. 6) for the first-mentioned calibration;it is connected to an amplifier Ao, supplying avoltage with an r.m .s. value of 300 V. This voltageis stabilised by a regulating device R as discussedabove; the components Ca, Dl' Cg, RI and R2

correspond with those correspondingly marked infig. la. The only difference between the two circuitsis the presence of an amplifier in the one underdiscussion, which makes no difference to the princi-ple, but as observed before, improves the stabilisationfactor. The reference voltage Eref is obtained froma stabilised direct voltage source of 310 V via apotentiometer.

Panel 5 (fig. 6) contains twelve oscillators, forthe frequencies 1, 5, 100 and 1000 kc/s and 1, 3,10, 15, 18, 20, 25 and 30 Mc/s, for checking thefrequency characteristic 8). The circuit voltage ofthese oscillators is stabilized at 30V (rm .s. value), byapplication of the regulating principle discussedabove.

Gt1601612:2-----1I

L___.~-4.."._~r--+-L~~~~--4J_r'_------~__j:L _.J

73181

Fig. 7. Circuit diagram for the calibration of the meter GM 6016and its attenuator Z2. 00 oscillator with frequency 100 kc/s.Ao amplifier. R, with the component parts Cg, Dl' Ca, RIand Rz: unit for the producing of a control voltage (sec fig. la)which stabilizes the output voltage of Au. The reference voltageErer is adjustable. Fo low-pass filter. Z, attenuator adjustablein calibrated steps.

Absolute calibration and checking of the auen iuuor at 100 kc]«

Although it actually falls outside the scope of this article,a brief outline of the calibration procedure is given below.

An attenuator Z. (fig. 7) belonging to the calibratinginstallation and to which 300V at 100 kc/s is applied, deliverscalibrating voltages of 300,100, 30 and 10 V to a series ofsockets, and in addition to this, by means of a switch, voltagesof 3 and 1 V and 300, lOO, 30, 10 and 3 IIIV can be obtained.

8) The presence of two osci Ilators for 1000 kc/s (= lMc/s)will be explained later.

310 PHILlPS TECHNICAL REVIEW VOL. 14, No. 10

Fig. 8. Device for adjusting the 100 kc/s input voltage of attenuator Zl exactly to 300 V.Switch S in position 1: After the correction potentiometer ReoT' has been set in accord-

ance with the temperature of the standard cell Nl, the direct voltage E is so adjustedthat the deflection of the galvanometer Cl becomes zero, E is then exactly 300 V. ThedeHection of the galvanometer Cz, which is connected to four thermocouples Th connectedin series (of which only one is shown in the diagram), is reduced to zero by means of thecurrent le, flowing through the resistor Rs.

Switch S in position 2: The reference voltage Ere[ of the oscillatOl"-amplifier Oo-Ao-R(see fig. 7) is so adjusted that Cz again returns to zero at exactly the sa me value of Jeas before. The r.m.s. value of the voltage at the input of Z, is then exactly 300 V.

First the calibrating voltage of 3 mV is applied direct tothc terminals of the meter GM 6016 which has to be checkcd;the meter is so corrected that it gives the proper deAcction.Next one changes over to the higher calibrating voltages,the capacitive attcnuator beJonging; lO the m e tcr (22' fig.7)being interconnected. This attenuator, which is continuouslyvariable, is each time adjusted in such a way that the metergives the correct indication; at these positions holes lUC drilled,so that a catching device, consisting of a steel ball, which ispressed into the hole by a spring, stops the at.tcu ua tor.

The question, however, arises how it can be ascertainedthat the high-frequency alternating voltage at thc input ofZ[ is exactly 300 V. This is checked by means of a directvoltage, by co mparison with the voltage of a standard cell.

The procedure is as follows.The input side of Zj is disconnected from the amplifier Ao

and connected to an adjustable direct voltage source E """300 V (fig. 8). To make E exactly 300 V it would of coursebe possible to use a moving-coil voltmeter, but greater accuracycan bc obtained by comparing a given fraction of E with thevoltage of a standard cell in a compensation circuit. Therefore,the direct voltage source is shunted by a potentiometer,consisting of the (substantially) fixed resistors R,>] and Rpz

(fig. 8). The ratio of these resiatances has been so selectedthat the voltage across Rpz is equal to that of the standardcell N, if E is exactly 300 V. To make E = 300 V, E is varieduntil the deflection of the galvanorneter Cl has become zero.

Since the voltage of a standard cell depends to a certaindegree on the temperature, the ratio RpI: Hpz has to be variableto a slight extent; the small correction potentiometer Rcon"

serves this purpose. This potentiometer is provided with atemperature scale; it is set to the tempera ture read on athermometer inserted in the box of the standard cell (Philips

type GM 4.569, fig. 9).If a direct voltage of exactly 300 V at the input of the

attenuator Z[ has tb us been obtained, a ccrtain current wiJlflow through each of the resistors of ZI. One of these currentsis measured in relative value by means of four thermo-couples Th. connected in series (four to increase the sensitivity )

z,

73182

Fig. 9. Close-up view of the panels 2, 3 and 4.Cf. fig. 8.: N standard cell (type GM 4,569) with thermometer.

C galvanometer serving either as Cl or as Gz. S switch changingover from direct voltage to alternating voltage. Z[ attenuatoradjustable in steps. B socket to which the attennator of themeter GM 6016 is plugged.

Cf. fig. 10: SI' S2' S3 frequency selector switches. Pl-P1Zpotentiometers for adjusting the oscillator voltages to thecorrect value. K2 coaxial cable. H probe with germanium diodeand capacitive voltage divider (Cez and C,-CZ in fig. 10).The meter GM 6016 to be checked is connected to the plugJ. M meter for the comparing the oscillator voltages.

·";==-~~-I-'--f) , e

APRIL 1953 OSCILLATOR WITH CONSTANT OUTPUT VOLTAGE 311

.-----,f<.. 1 1±;---o+7ft======1::::::;:::::::=======1 GM60f6:

1 1L- .J

and a galvanometer G2 (fig. 8). This too, is effected by a zeromethod, for the galvanometer' circuit includes a resistor R.which is traversed by an adjustable auxiliary directcurrent le. This current is adjusted in such a way that thedeflection 'of the galvanometer G2 becomes' zero, i.e. that thevoltage across R. exactly compensates the output voltageof the thermocouples; le then is a measure of the' currentflowing through the heaters of the thermocouples. Now Zl isswitched over from the direct voltage source to the amplifierAo, which supplies an alternating voltage of 100 kc/s, andthe reference voltage Eref is adjusted until the deflection ofthe galvanometer G2 again becomes zero. In that case thealternating voltage at the input of Zl has an r.m.s, valueof exactly 300 V (0.1% difference already makes G2 deflect15 gradations), and at the output terminals of Zl the cali-brating voltages 300 V, 100 V, 30V, ,..3 mV are available.The attenuator Zl has been corrected for stray capacitances,so that the attenuation is the same with an alternatingvoltage of 100 kc/s as with a direct voltage.The deflection of the meter GM 6016 depends on the form

factor of the input voltage; the meter, however, has beencalibrated in r.m.s, values. The calibration is correct only forsinusoidal voltages. Although the output voltage of theamplifier Ao shows only little distortion, a simple low-passfilter [t'o has been plaeed behind this amplifier; in this waythe distortion of the voltage across the meter GM 6016 is keptlower than 0.1%. '

Checking the frequency characteristic

Via a coupling coil a voltage of 300 mV is obtained fromeach of the twelve oscillators of panel 5. These coupling coilsare connected to two selector switches (SI and S2' fig. 10) onpanel 4 by coaxial cables; with these switches a measuring

9'IIII

04

provided with a sliding core of Ferroxeube; see the article'referred to in note 7), page' 209).

Each of the twelve oscillators is provided with a potentio-meter (Pl'" P12, fig. 9) with which the reference voltage andconsequently the output voltages can be varied.The way in which the equality of the measuring voltages

is checked, requires' further comment. A direct voltage isderived from the alternating voltage by means of a germaniumdiode Gel' A vibrating contact (S4' fig. 10) converts this intoa low-frequency pulsating voltage which is amplified by theamplifier Al and measured in a relative measure by the meterM. At the lower frequencies (1 to 1000 ke/s), from which achoice, can be made by means of the selector switch Slo acathode follower is incorporated behind SI' The germaniumdiode is connected direct to this cathode follower, and themeter GM6016, which has to be checked, via a coaxial cable.However, at frequencies exceeding 1 Mc/s the differencebetween the voltages at the beginning and the end of thecable would become excessive (see the article mentioned innote 7), page 2140);therefore, at these higher frequencies, useis made of a germanium diode Ge2mounted in a probe at thesame end of the cable where the meter is connected.The frequency 1000 kc/s = 1 Mc/s occurs both in the lower

and in the higher frequency range, so as to ensure a goodagreement between the measurements in the two ranges. .By means of an attenuator a voltage of 3 mV is derived from .

the voltage of 300 mV which is induced in each of the couplingcoils, for checking the meter without using the attenuatorZ2belonging to it. For the frequencies from 1 to 1000 kc/s thefirst-mentioned attenuator consists of resistors (rl>72, fig. 10),which are contained in the measuring rack; for the frequencies1 to 30Mc/sit consists of capacitors (Cl' C2), which are mountedin the probe ~t the end of cable K2.

GM601

Fig. 10. Set-up for checking the ffeqeuncy characteristic ofthe millivoltmeter GM6016.01-012 oscillators with automatically stabilised voltages, with frequencies 1 kc/s to 30Mc/s.SI selector switch 1-1000kc/so S2selector switch 1-30Mc/s. S3selector switch (1-1000 kc/s)/(1-30 Mc/s). T cathode follower. rI' r2 and Cl' C2 voltage dividers, by which not only300 mV but also 3'mV are available. Kl> K2' coaxial cables. Gel>GC2 rectifying circuits'with germanium diode (ip.the form of a probe). S4vibrating contact. Al amplifier. M meter.

voltage in either frequency range 1-1000 kcts or 1-30 Mc/sis selected. With equal measuring voltages, the meter GM6016should give the same deflection within certain tolerances;should it fail to do so, then certain components of the metershould be readjusted until the frequency characteristic hasbecome sufficiently flat (to facilitate this, certain coils are

73183

Summary: In accord~nce with a well-known principle ofregulating technique, the output voltage of a valve oscillatormay be stabilised by the mutual comparison of two currents,one of which is obtained by rectifying the output voltage,whereas the other originates from a reference voltage source;,from the difference between these currents a regulating voltageis derived, which serves for biasing the oscillator. valve.

,312 PHILIPS TECHNICAL REVIEW,-.._"J

VOL. 14, No. 10

ABSTRACTS OF RECENT SCIENTIFIC PUBLICATIONS OF THEN.V. PHILIPS' GLOEILAMPENFABRIEKEN

Reprints of these papers not marked with an asterisk * can be obtained free of chargeupon application to the X~~~"\i)tN:,"'(I}tjNOCOO~ Administration of the ResearchLaborntory, Kastanjelaan, Eindhoven, ~etherlands.

, In this way, the magnitude ofthe output v~ltage can, with theaid of only a few extra circuit elements, be made highly insensi-tive to mains voltage fluctuations and to changes in the mutualconductance of the valve and of the impedance of the oscil-latory circuit. The output voltage is less distorted than withan ordinary Le oscillator.By making the reference voltage controllable, it becomes

possible to adjust the output voltage to the value required,and by superimposing an alternating voltage on the referencevoltage, the amplitude of the output voltage can he modulated.The upper Iimit of the modulation frequency and that of themodulation depth are discussed.The tendency towards squegging can be counteracted by

2012: R. van del' Veen: Fluorescence andinduction phenomena in photosynthesis(Physiologia plantarum 4, 486-494, 1951).

Fluorescence of chlorophyll in leaves showsinduction phenomena during the first few minutesof illumination, which are very much like those ofphotosynthesis. The influence of preceding darkperiod, of temperature, of light intensity and ofCO2 and O2 pressure on the adaptation curve wasstudied. An increase of photosynthesis duringadaptation is nearly always correlated with adecrease of fluorescence.The only exceptions are: (a) in the temperature

limited range of photosynthesis, a lower tempera-ture causes less photosynthesis, but does not resultin any marked change in fluorescence and b) ahigh O2pressure causes both a lower rate of photo-synthesis and lower fluorescence.

The results of these experiments are thereforemostly ïn agreement with the view of Kat z; thefluorescence of chlorophyll can often be used as aflow meter for the energy of photosynthesis.Sometimes, however, photosynthesis and fluores-cence do not show any correlation. In these casesit is probable that the light energy is used for thesplitting of H20 but that afterwards. back reactionscause a reduction of the photosynthesis.

2013: W. de Groot: On the definition of standardilluminant A [Physica 17, 920-922, 1951,No.10).

In the Stockholm meeting 1951, the C. I.E.provisionally defined standard illuminant A by the

giving the correct value to ~ertain capacitances. For theoscillator to start smoothly after switching on, care shouldbe taken to prevent the current supplied by the referencevoltage source from flowing as control grid current throughthe valve; to this end an auxiliary diode can be inserted inthe circuit. -The valve EBF 80 contains u pentode system, which may

serve as an oscillator valve, and two diode systems, one ofwhich can ,rectify the alternating voltage to be stabilized, whilethe other may function as the auxiliary diode just mentioned.A calibrating system for the type GM6016 electronic milli-

voltmeter (1000 cJs to 30McJs)is discussed as a practical appli-cation.

value {J = c2/T = 5.0386 (= 14350/2848 ~ 14380/2854) to be used in Planck 's (orWien' s) radiationformula. It is pointed out that if one wishes torealise the standard (in ~ien 's approximation) byconstructing a black body, the monochromaticpower ratio of which (with respect to a black bodyat the melting point of gold) is exp (a/A), wherea = C2/TAu -{J, a change in C2 and TAu in generalinfluences a, if {J is kept constant. The spectraldistribution of this source will in general deviatefrom the calculated value unless the "true" valuesof C2 and TAu are used in calculating a.

2014: A. Bierman: A systematic method with newstandards to determine the correct magnitudeof radiographical exposures (Acta radiologica36, 311-323, Oct. 1951).

The value of a radiographical exposure E isdetermined by E = (kV/I00)5 mA·sec/m2 (in whatthe writer calls "radiographic exposure units" or"1' e u"). A standard object (a water phantomwith definite dimensions) is used to' characterisethe radiographical outfit with the "specific exposure"Es. This is the value of E required to obtain a filmdensity of unity when radiographing the standardobject. The exposure required for a certain objectca~ he computed with the aid of Es and "relativeexposure" E/Es. The values of of E/Es appropriateto the various objects constitute a "basic exposuretable", from which the correct exposure is derivedby means of the formula E = (E/Es) X Es. Allcomputations can be avoided by the use of twonomograms.