frequency-analog speed control - philips bound... · frequency-analog speed control ... used...

260 Philips tech. Rev. 33, 260-271,1973, No. 8/9

Frequency-analog speed control

H. Kalis and J. Lemmrich

GeneralAn electronic method of control that has been known

since the early thirties is that of the phase-lock-looptechnique. In this method the phase angle of a periodicsignal whose frequency corresponds to the value of thequantity to be controlled is continuously comparedwith the phase angle of an applied signal whose fre-quency represents the specified value. A change in thephase difference between the two produces an errorsignal that counteracts the change; in this way the twofrequencies are permanently synchronized and thequantity to be controlled is kept to the required value.A control system of this type, in which the quantitiesare represented in the form of a frequency, is referredto as frequency-analog control at Philips Forschungs-laboratorium Hamburg [11. The principle of frequencycontrol is convenient for the control of many kinds ofquantity, but is particularly suitable for controlling thespeed of electric motors, where the quantity to becontrolled is essentially periodic in nature.

We shall start by way of introduetion with such asystem; see the diagram of fig. J. A voltage Vrer ofsinusoidal waveform and angular frequency Wrer (orthe sum of a number of signals of different frequencies)is multiplied by a voltage of sinusoidal waveform andangular frequency W in a multiplier circuit M. Theoutput voltage of the multiplier is applied through alowpass filterLP to an oscillator VCO whose frequencycan be controlled by a voltage; this in turn supplies thevoltage V of angular frequency w.

The multiplication produces a voltage Va at the out-put of the multiplier, given by:

Vo=-r ABC{cos(Wrert wt-cp)- cos (Wrert + wt+ cp)}.

Here A and B are the amplitudes of the two voltages,C is the multiplication constant and cp is the phase ofthe voltage V with respect to the reference voltageVrer. The output signalof the multiplier thus containsthe sum and difference frequencies of the input signals.

The lowpass filter suppresses all frequencies abovethe cut-off frequency wc/2n. It always suppresses thesum frequency, and also the difference frequency if thisis above we/2n. If however the frequency of the con-trolled oscillator is set close to the reference frequency

Dipl.-Ing. H. Ka/is and Dipl.-Ing. J. Lemmricb are with PhilipsForschungslaboratorium Hamburg GmbH, Hamburg, West Ger-many.

Wrer, e.g. by the auxiliary voltage VI, so that thedifference IWrer - W I is less than Wc, then synchroniza-tion is obtained automatically and W becomes equal toWrer. Synchronization is initiated because the controlsignal Va increases if the difference frequency becomessmaller; the frequency of the controlled oscillator istherefore pulled in to the reference frequency. Once thesystem is synchronized the multiplier circuit acts as arectifier, controlled in phase by the oscillator, whichtakes a phase difference such that

Wrer = k{tABCcos(- cp)+ Vl}, (1)

where k is a generator constant.It can be seen from the diagram of fig. 1 that this

method of obtaining synchronization depends on theexistence of a control loop. The error signal in thisloop is given by the phase difference cpbetween the twoinput signals while the angular frequency W is thecontrolled quantity [21.

Now let us consider a phase-lock-loop system inwhich the voltage-controlled oscillator VCO is replacedby a motor whose speed of rotation is determined bythe voltage Va. The shaft of the motor is assumed tobe connected to an a.c. tachogenerator that gives anoutput signal at a frequency proportional to the angularvelocity w corresponding to the speed of rotation;this signal corresponds to the voltage V in fig. 1. Withsuch a system there could be no possible error betweenthe speed of the shaft - represented by the angularfrequency to - and the specified value Wrer: the speedcontrol would therefore be ideal. The idea has beenput into practice in various places in the last ten yearsor so [31, but had been used as early as 1938 [41.

V,ef=A sin Wreft

Fig. 1. Phase-lock loop for synchronizing the frequency of a volt-agecontrolled oscillator VCO with a reference frequencywrer/2:n:.The output voltage V of the oscillator is multiplied by thereference signal Vref in the multiplier circuit M and the resultantsignal is passed through a lowpass filter LP to give the voltage V•.This voltage Va. controls the oscillator. An auxiliary voltage VIcan be applied externally to initiate the synchronization.

Philips tech. Rev. 33, No. 8/9 FREQUENCY-ANALOG SPEED CONTROL 261

Phase-lock loops for speed control of electric motors

In this section we shalliook at the system of fig. 1 tosee if it is the most suitable one for speed control andto find the best way of replacing the controlled oscil-lator by a motor fitted with a sensing device. We shallalso attempt to illustrate the most important commonfeatures of the various versions of the system producedin our Hamburg laboratories by two simple but typicalexamples.

Since the measured frequencies have to be producedby tachogenerators, the frequencies used in motorcontrol are relatively low (a few hundred hertz). Thesignals are usually rectangular pulses, because theseare the easiest to produce. Rectangular pulses areused without exception in all the literature that wequote.

Control with the aid of analog signals

In our first example of a frequency-analog speedcontrol the multiplier of fig. I is replaced by a phase-detector circuit in which the phase of the pulses is con- gverted into a voltage.

The operation is explained with the aid of fig. 2. Fig. 2a showsthe reference signal (corresponding to Vrer) and fig. 2b the meas-ured signal (corresponding to V) as a function of time; a phasechange in the measured signal is produced at time to, for exampleby changing the load on the motor. Fig. 2c shows a sawtoothvoltage whose slope angle IX is proportional to the referenceangular frequency Wrer; the start of the sawtooth always occursat the leading edge of the reference-signal pulse. The phase dif- ,ference between the reference and measured pulses is convertedinto a quasi-steady-state signal V.(t) by a hold circuit, whichholds the instantaneous value of the sawtooth voltage at theinstant when a measured pulse appears and produces this voltageat the output as a control signal (fig. 2d). This control signal isindependent of the frequency, since the slope angle IX of thesawtooth voltage is proportional to the reference frequency.

Fig. 3 shows the complete system, including themotor M, in the form of a block diagram. It is assumedthat M is a d.c. motor with separate excitation because

[1] D. Gossel, Mef3systeme und Regelungen mit Frequenzsigna-len, Messen, Steuern, Regeln 14, 22-28, 1971.D. Gossel, Frequenzanalogie - Ein Konzept für Mef3- undRegelsysteme mit digitaler Signalverarbeitung, Elektrotechn.Z. A 93,577-581, 1972.

[2] A general treatment of phase-lock loops is given in: F. M.Gardner, Phaselock techniques, Wiley, New York 1966.

[3] M. J. Campos Costa, Obtaining precise induction motorspeed, Control Engng. 10, No. 6, 92-93, 1963.J. Lemmrich, Der synchronisierte Induktionsmotor, Elek-trotechn. Z. A 85,724-726, 1964.E. K. Howell, Solid-state control for dc motors providesvariable speed with synchronous-motor performance, IEEETrans.IGA-2, 132-136, 1966. ',H. Kalis and J. Lemmrich, Frequenzanaloge Drehzahlrege-lungen I, 11, Regelungstechnik 16, 497-502, 555-562, 1968.

[4] H. Rinia, Television with Nipkow disc and interlaced scan-ning, Philips tech. Rev. 3, 285-291, 1938.

of the extremely convenient static and dynamic equa-tions relating rotor current and motor torque, and alsomotor torque and speed.The part of the system corresponding to the voltage-

controlled oscillator of fig. 1 - motor, controlledrectifier bridge (CRB) and pulse tachogenerator (L,SD, Pho, PS) - is enclosed by a chain-dotted line.Since there are time delays present inside this' block

Q ~ ~ ~ ~ ~I I I I II I I I II

~

I

~

I

~

I

~

II I I I II I I I I

Q II I I II II I I I I I I

I I I I I I I I II I I

Fig. 2. Conversion of the. pulse phase to a voltage. a) Referencepulses. b) Measured pulses. c)· Sawtooth voltage whose startalways coincides with the leading edge of the reference pulse.When the measured pulse arrives the sawtooth voltage hasreached a value proportional to the phase difference betweenreference and measured pulses. d) This value is held in a holdcircuit. At time to there is a step in the phase difference, e.g. as aresult of a sudden change in the load on the motor.

,------,I II I

J--'-- III

+

r------------I, "-' CRB

Fig. 3. Frequency-analog control of a d.c. motor M with thespeed of rotation measured by a light source L, a slotted disc SDand a photodetector PIlO. The rotor current is controlled by acontrolled rectifier bridge CRB. This circuit is controlled by theanalog output signalof the phase detector and synchronizingcircuit PD + Sy, after subtraction of the damping signal Vrn,which is proportional to the speed and is derived from the photo-detector via the pulse shaper PS and the frequency/voltageconverter.

262 H. KALIS and J. LEMMRICH Philips tech. Rev. 33, No. 8/9

, ,--- .--------------

(the moment of-inertia of the rotor and time constantof the rotor circuit) its dynamic behaviour does notcorrespond to that of the oscillator of fig. 1; a dampingsignal is required, to prevent instability. This is suppliedby a circuit that converts frequency to voltage (the partof the circuit inside the dashed line in fig. 3). More willbe said about this circuit later.

The operation of the speed-control system is asfollows (leaving aside for the moment the signal fromthe frequency/voltage converter). The output voltagefrom the phase detector PD, whose operation has

wref

it also follows that for steady-state conditions

Tl = k!k2I1cp,

where k2 IS another gain factor. The phase differencebetween reference and measured signal is thereforeproportional to the load, from which it can be seenthat this - and any other frequency-analog-controlledmachine - behaves like a synchronous machine [5].

To bring out these relationships more clearly, weshall show how the complete system reacts to a suddenstep in the load when it is in equilibrium. After the

r-------..,J II I

II II II W IL J

Fig. 4. The block diagram of fig. 3 redrawn in control-theory terms. The transfer function ofeach block is shown; s is the complex frequency a + jw.

already been described, functions as a reference valuefor the rotor-current control circuit of the d.c. motor.This ensures that when the very short-lived transientsfrom the rotor-current control circuit have died awaythe torque Te of the motor is directly proportional toVa. The relation between the angular velocity meas-ured by the sensing device and the torque Te of themotor is given by the equation:

Te - Tl = I dwrfdt = I d2cpr/dt2, (2)

where Tl is the load torque, I the total moment ofinertia of the rotating components, Wr their angularvelocity and cpr the instantaneous value of the angularposition of the rotor. The sensing device gives a meas-ured frequency

w/2n = kWr/2n,

where k is the number of pulses per revolution.If the phase-lock loop is in equilibrium (Wrer= co),

then dWr/dt = 0 and it follows from (2) that Te is equalto Tl and therefore, because of the current controlmentioned above, equal to Vak! (k! is a gain factor).Since the reference voltage Va originates from the phasedetector and is directly proportional to the phase dif-ference I1cpbetween the reference and measured pulses,

increase in the load the motor slows down because themotor torque Te is now too small. This immediatelycauses a progressive increase in the phase differencebetween Wrer and co, and hence an increase in Va. Thisprocess continues until the voltage Va is large enoughto ensure a sufficiently large motor torque. The newequilibrium has then been reached.

(3)

To demonstrate the need for the damping by means of thefrequency/voltage converter, we represent the circuit shown inthe block diagram of fig. 3 in control-engineering terms (fig· 4).The appropriate transfer function is shown in each block; theindividual transfer functions can be combined to give the transferfunction G of the complete control loop. The independent vari-able in these functions is the complex frequency s = a + jw,used as an operator in control and circuit theory. The sameblocks as in fig. 3 are enclosed by chain-dotted and dashed lines.It can be shown from fig. 4 that for k3 = 0 the transfer functionfor the open loop is given by

(4)

where Tr is the time constant of the rotor circuit. A con trol loopwith such a transfer function is always unstable when closed.If k3 is made greater than zero the loop becomes stable. Thetransfer function of the open loop is then:

kkl k2(1 + k3s/k2)G = 1s(1 + STr) S

(5)


For every combination of parameters there is a value of k3 thatwill give critical damping of the transient effects.

Apart from the quantization that arises because reference andmeasured values are only compared at discrete times, the transferfunction ofthis frequency-analog control system is identical withthat of a conventional PI control for the speed. (In which thecontrol is provided by a signal P proportional to the speed and asignal I proportional to the integral of the speed.)

Up to this point we have been concerned with main-taining the equality COrer = w. However, at the momentwhen a stationary motor is switched on, 00 = 0 andCOrer> O. The phase-lock loop is therefore not inequilibrium. It has to be brought into equilibrium bythe synchronizing circuit Sy (fig. 3). This is essentiallya f~equency-comparison circuit; if COrer =1= COit producesthe required acceleration or deceleration of the motor,and when COrer has become equal to co,Sy switches overto the phase comparison described earlier.The function of Sy can be carried out by various

kinds of circuit. A very simple circuit that makes useof the phase detector described above (fig. 2) is a five-stage ring counter - a shift register with the outputconnected to the input - that can count both back-wards and forwards (i.e. the pulses can progressthrough the shift register in either direction). The ringcounter used here has automatic latching: in the finalstate '5' it suppresses all forward-travelling pulses andin the final state '1' it suppresses all backward-travellingpulses. The reference pulses travel forwards, the meas-ured pulses backwards.

If COrer> 00, then on average there are more referencepulses per unit time than measured pulses and thecounter is switched erratically back and forth between'4' and '5'. If COrer < co,then it is switched erraticallybetween states' l' and '2'. This switching back and forthcan be used in both cases to make the control unit -e.g. a controlled rectifier - give either maximumacceleration of the motor (for '5', '4'), or maximumdeceleration (for '1', '2'). When the two frequencieshave been made equal by this synchronization process,a characteristic pattern results. Depending on the signof the previous frequency deviation, either two for-ward pulses appear between two backward pulses (forswitching between '1' and '2'), or two backward pulsesappear between two forward ones (switching between'4' and '5'). When this indication is received the syn-chronizing circuit switches to the phase detector: for-ward and backward pulses now appear singly butalternately, which shows that the ring counter is switch-ing between '2' and '3' or '3' and '4'. The phase dif-ference between forward and backward pulses is nowconverted with the aid of the phase detector as de-scribed earlier into (say) a negative control voltage forbraking and a positive one for acceleration [61.

We shall now end our description ofthis first exampleof a frequency-analog motor control with a short dis-cussion of the frequency/voltage converter and. thesensing device.The most usual way of converting a frequency into

a voltage is by deriving the average of the pulse fre-quencies with the aid of a lowpass filter, but thisprocess givestoo large a delay to permit effectivedamp-ing. Only a very fast process making use of a holdcircuit is of any use; the operation is shown in fig. 5.The leading (or trailing) edge of each measured pulse

fmin .

AIJ:~1~~-~-f

Fig. 5. Conversion of the measured frequency into a voltageVmens. At a time tmln after a measured pulse a function generatorstarts to generate the function Alt (the time t is related to theleading or trailing edge of each pulse), and continues to operateuntil the next measured pulse arrives. The time between two pulsesis equal to T, so that the voltage at that instant is AlT; this valueis retained in a hold circuit, and since it is inversely proportionalto T it is proportional to the repetition rate of the measuredpulses.

starts an electronic function generator, which waits aconstant time tmin and then generates a voltage curveof the shape Vet) = Alt; Vet < tmin) = A/tmin. On thearrival of the next measured pulse, at time t = T, theinstantaneous value of the voltage is held in a holdcircuit and the whole procedure starts again. In thisway a measured value Vmeas = A/T can be obtained,where T is the period of the measured frequency. Theapplication of this function generator thus enables avoltage proportional to 00 to be derived for each pulseperiod.Next we come to the sensor. Fig. 3 shows a well

known type of sensor - a disc with equidistant slots.The disc is attached to the motor shaft; the frequencyis obtained with the aid of a lamp and a photosensitivedevice connected to a pulse shaper. However, we havefound it preferable to use a carrier-frequency technique,based on a new principle, in which single-sidebandmodulation is obtained from 'mechanical' signals.This can be done by using small rotary-field devices(synchros) driven by the drive shaft and with the car-rier signalof frequency wo/2n supplied to the electrical

[5] See: E. M. H. Kamerbeek, Electric motors, this issue, p. 215.[6] A system opèrating in a closely similar way but for another

application and in only one direction has been described indetail by P. Diekmann in: Digitale Frequenz- und Phasen-vergleichsschaltung für nachgesteuerte Oszillatoren, Int.elektron. Rdsch. 24, 231-232, 1970.


input (the stator or rotor). The signal obtained at theoutput then has either the sum frequency (wo + wl')/2nor the difference frequency (wo - wr)/2n depending onthe sense of the mechanical rotation with respect to therotary field.This arrangement does not have the disadvantage of

the slotted-disc sensor (and any other sensor that givespulses at a repetition rate proportional to the shaftspeed) that the scanning rate of such a discrete systemis so low that this quantization limits performance andcan introduce instability if there is a large decrease inspeed. Our carrier-frequency technique also avoidsanother difficulty that can occur with the more usualdouble-sideband modulators. These produce twosidebands, Wo + wand Wo - w; at low angular fre-quencies w the two sidebands are very close and can-not be separated by using a linear filter.

Purely digital control

We now come to our second typical example of afrequency-analog speed-control system. The essentialdifference between this system and the first one is thathere there is no analog transfer in the actual controlloop (fig. 6), only switching. This approach makes thesystem very insensitive to interference.As in fig. 3 the block PD + Sy (phase detector and

synchronization) represents a bi-directional counterwith almost identical control; but in this example of asingle-quadrant drive - in which only one sense ofrotation and motor operation are required - thereare four states, not five, with latching in states '1'and '4'. The states of the counter have a differentsignificanee here. In the states '4' and '3' the amplifierPA, which amplifies the trigger pulses for the controlledrectifier, is gated off; while the counter has state' l' or'2' the thyristors Th: and Th2 are triggered contin-uously. The reference pulses are derived from the zerocrossings of the mains waveform by a pulse shaperPSI: the mains voltage acts here as the reference signal.The measured pulses are derived from the zero cross-ings of the output voltage of the synchro generator bya pulse shaper PS2. For the present we shall leave asidethe functions of the phase modulator PhM connectedin the signal path and the/IV converter for damping;their operation will be more fully explained later. Thespecified speed is presented here in the form of thecarrier frequency for the synchro. The frequency rangeto be covered is given by

wm ~ Wo ~ Wm + 2nnmax,

where Wm is the angular frequency of the supply mainsand nmax is the maximum speed of the motor (inrevolutions per second). The angular frequency of thesynchro output voltage is equal to Wo- Wr.

PD+Syw(f-Llf)

Fig. 6. Speed control for a d.c. motor. M motor. S synchrogenerator for generating the measured frequency. G three-phasegenerator. PS1,2 pulse shapers. PhM phase modulator. PA pulseamplifier.

After the system has been switched on, wo- Wl'> wm,

so that the ring counter switches erratically back andforth between states' l' and '2' because of the countingdirection, as discussed earlier. Continuing with the con-vention adapted above the thyristors are then per-manently triggered and the rotor voltage correspondsto the average of the complete half-cycle of the 50 Hza.c. voltage. The maximum motor torque is then de-livered and the motor consequently runs up to speed.This makes w = Wo - Wl' smaller and when thesituation Wm = W is reached state '3' of the counterappears for the first time. Since state '3' is initiated bythe reference pulse, then in our convention the trigger-pulse amplifier is gated off at the zero crossings of themains voltage, so that triggering does not start at thebeginning of the next half-cycle. Triggering does notstart until the measured pulse switches the counter backto state '2'. From now on state '2' and state '3' areselected alternately, which has the effect of phasecontrol of the mains voltage and also means thatsynchronization has been achieved. Fig. 7a shows therotor voltage derived from the mains voltage, fig. 7bshows the output voltage of the synchro, fig. 7c thereference pulses derived from the mains voltage andfig. 7d the measured pulses derived from the synchrovoltage. The hatching indicates the part of the cycle inwhich each thyristor conducts..If the load is increased the machine starts to runmore slowly. Since W = Wo- Wr, this .results in adecrease in the phase difference between wand Wm andan increase in the firing angle (x. The motor torquetherefore increases and the control takes up a new


equilibrium. This is very similar to what happenedin the first example for an increase in load, but onedifference here is that the ring counter is involved inboth measurement and control.

To explain the function of the phase modulator PhM and thefrequency/voltage converter, let us look again at fig. 7. We cansee at once that an intentional change in the phase differencebetween OOm and 00 will have an immediate effect on the motortorque because of the phase control. In the same way as in thefirst example damping can be provided by introducing a phasechange proportional to oo. The block diagrams of the control aretherefore identical for the two systems.

The phase modulator consists of a simple monostabie circuitwhose delay t!.t can be controlled by a voltage. The control is gtaken from the output voltage of an 1/ V converter, like the onedescribed earlier, via a highpass filter. The original value of thedelay time of the monostabie circuit, the 'operating point' forthe control, has no steady-state significanee since it merely pro-duces a constant angular difference. It can in fact be used toapply angular corrections to a rotating motor with a controlvoltage. The dynamic operation is of course slightly affected bythe small extra delay time.

Two other versions

To end this section we shall take a quick look at twofrequency-analog control systems for induction mo-tors, which are widely used in drive systems. The firstsystem is for a slip-ring armature motor, the secondfor a squirrel-cage motor.

Since the slip-ring armature motor is in itself anelectromechanical frequency-difference device, like asynchro, no sensor is necessary, and the frequency ofthe rotor current will serve as the measured quantityrelated to the speed (seefig. 8). To obtain the desiredscanning speed the rotor-current frequency ismultipliedby six, making use of the three rotor phases. In thisway a measured-frequency range of 300-50 Hz corre-sponds to speeds of 0 to about 21 tev]« in a four-polemachine. This same frequency range also has to becovered by the reference generator G [7]. Control of themotor torque is obtained with the aid of a d.c. chopperCh and a full-wave three-phase rectifier that convertsthe rotor current to d.c.; this d.c. current, which alsoflows through an external rotor resistance, is period-ically interrupted by the chopper. The ratio of thetime switched on to the time switched off (mark/spaceratio) determines the effective resistance, giving rapidelectronic control of the motor torque [5] [81. As beforedamping is provided by phase modulation of thereference frequency.The motor torque and speed of a squirrel-cage motor

can be controlled by using an inverter; a voltage-controlled oscillator is then required that will controlthe frequency of the output voltage from the inverter,and also its amplitude, to obtain as constant a flux aspossible in the machine; seefig. 9. The phase detector

a

I I II

I I II I I I

~I n rII I

I II I

I I

~___fl n-- .. t

Fig. 7. a) Rotor voltage of the d.c. motor in the control systemoffig. 6. At any given moment one of the two thyristors will beconducting (firing angle ex); it will remain conducting after themains voltage has changed sign since the energy stored in themagnetic field of the rotor still has to be discharged. Thisthyristor remains in the conducting state until the other one istriggered. b) Output voltage of the synchro, c) Reference pulsesderived from the mains voltage. d) Measured pulses derived fromthe output voltage of the synchro; these determine the firingangle exof the thyristors. .

3rv

W

WChwref

Fig. 8. Frequency-analog control of a slip-ring armature motorby measurement of the rotor frequency. This is multiplied bysix and compared with the reference frequency produced by thegenerator G; the error signal resulting from this comparisoncontrols a d.c, chopper Ch. This chopper interrupts the rotorcurrent successively, after full-wave rectification in a three-phaserectifier, with a variable mark/space ratio and connects them toa resistive load. This varies the effective resistance in the rotorcircuit and hence the motor torque.

and the synchronization stage operate in the same wayas in the first example given in this section, while themeasurement is made as in the second example, butwith the aid of the upper sideband.

Since the flux in the machine is taken to be constant,

[7] J. Lemmrich, Frequenzanaloge Motorsteuerung mit kontakt-losen Bauelementen, Elektr. Ausrüstung 1965, No. 2, 58-61.

[8] H. Kalis and J. Lemrnrich, D.C. chopper with high switchingreliability and without limitation ofthe adjustable mark/spaceratio, lEE Conf. Publn. No. 53, Part 1: Power thyristors andtheir applications, pp. 208-215, 1969.


HP

Fig. 9. Frequency-analog control of a squirrel-cage motor M byvarying the frequency of the supply with the aid of an inverter Ill.The output frequency and amplitude of the inverter are deter-mined by a voltage-controlled oscillator VCO. A synchro Smeasures the speed. HP highpass filter.

a larger motor torque can only be obtained at greaterslip, because the torque of a squirrel-cage motor isapproximately proportional to the slip in the operatingregion of the characteristic [sJ. This proportionalityrequires an extra control ofthe supply frequency, whichis obtained by adding to the control signal S2 a signalSI = k30Jr proportional to the speed. The coefficient k3

is given a value such that the rotating motor field is asclose as possible to synchronism with the shaft atS2 = O. Damping is derived from the signal obtainedthrough a highpass filter HP and the followingamplifier.

Many more examples could be listed, such as thecontrol of an induction motor by voltage chopping [9J.

However, we hope that the explanations given abovewill allow such combinations to be worked out inde-pendently.

Advantages of frequency-analog controlover conven-tional systems

Conventional systems of motor control are eithercompletely analog or partly digital. A group that arecompletely analog are the PI andPID control systems;these use control signals proportional to the speed error(P), to its time integral (1) and in some cases to the timederivative (D). The most important advantages offrequency-analog controls over these analog systemsare:

- Perfectangularprecision, allowing accurate tracking.- The angular position of drive shafts can be adjustedat will, even with the motor running, merely by chang-ing the phase of reference or measured pulses.- Much better signal/noise ratio (the information isnot contained in the amplitude).- Compatibility with electronic data-handling equip-ment.

Compared with high-quality digital control systemscombined with a minor analog P-control loop, theadvantages of frequency-analog control are:- There is no need for the P-control tachogenerator.- For the same discrimination of the sensing devicethe amplification that can be obtained in the controlloop is much greater.- The control quantity is continuously variable; thisis not so in some conventional systems.

Very accurate angular tracking can be achieved byoperating two frequency-analog control systems fromthe same reference generator, and constant speed ratioscan be maintained by using frequency dividers and acommon reference generator. Extremely low shaftspeeds can be obtained, right down to zero. At zerospeed the drive reacts like a torsion spring. With thisarrangement tracking systems can therefore be runright up to speed synchronously, even from rest.

Drying-oven conveyor belt synchronized with a printingpress

We now give a description of a frequency-analogcontrol that we have developed for the conveyor beltof a drying oven. The frequency signals used not onlyprovide the synchronization for the two drives, theyare also applied for other control purposes.

Many commodities are packed nowadays in tins orin glass bottles or jars with closures, such as screw-capsor 'crown corks' made from tinplate. In the manufac-ture of the tins and closures sheets of tinplate about1 m2 in area are printed during one revolution of thecylinder of a printing press and then carried alongbetween two carriers attached to the conveyor belt ofa drying oven. Fig. la shows a diagram of such aprinting-and-drying unit. The drive for the conveyorbelt has to meet the following requirements:1. The conveyor belt and the printing press mustoperate in synchronism.2. The speed ofthe belt should be continuously variablein a given range (e.g. 1 : 3).3. The belt must never become so slack that the carrier

[9] A. Walraven, Controlling the speed of small induction motorsby means of thyristors, Philips tech. Rev. 28, 1-12, 1967.K. Rennicke, Speed control of capacitor motors by varyingthe supply voltage, to appear in the next issue on electricmotors.

Philips tech. Rev. 33, No. 8/9 FREQUENCY-ANALOG SPEED CONTROL

st Pr

frames tip over. Also, the belt is so long that two drivingmotors are required.4. The phase angle between the cylinder of the pressand the separate carriers at the instant of transfer mustbe variable in steps or continuously, to enable differentsizes of sheet to be loaded correctly into the dryingsection.5. The time required to synchronize the conveyor beltto the selected phase must be less than 3 seconds. Thisis necessary for compatibility with other operations.

Previously all these requirements have been met byusing mechanical devices such as shafts and clutches.However, there are limits to the successful use of suchdevices when the process is speeded up or the instal-lation made larger. Electronic elements then have to bebrought in. The solution to the problem is split intotwo parts: driving the conveyor belt to match a controlvalue provided independently or by a sensor on theprinting press, and the phase adjustment and syn-chronization at the correct phase angle.

Driving the conveyor belt

The conveyor belt is driven at each end by a con-trolled induction motor with gearing. Since the upperhalf of the belt is in tension, MI (fig. 10) has to supplythe greater part of the total torque required to over-

Or

267

Fig. 10. The printing-and-dryingunit for tinplate. St stack ofsheets of tinplate to be printed.Pr: printing press. Dr dryingoven, with its conveyor beIt andupright carriers to carry theprinted tinplate through theoven. MI,2,a motors.

come the friction, while M2 determines the speed atthe point where the tinplate sheets are transferred bysupplying a greater or lesser additional torque. Toobtain good dynamic operation and a satisfactorystatic stituation, it is found desirable to arrange that aconstant torque, unaffected by the control and just toosmall to move the belt, is supplied by MI; the additionaltorque required to operate the belt, and dependent onthe load, is shared between the two motors. In thisarrangement the ratio in which the additional torqueis divided between the two machines is independent ofspeed and load, - a desirable condition for gooddynamic operation.

The speed of the slip-ring armature motor M2 iscontrolled by a frequency-analog system (fig. 11). Themeasured frequency is derived from the electrical rotorsignal (as explained on p. 265). The reference frequencyis produced in the rotor of a synchro S or by an elec-tronie frequency generator G. Comparison of the re-ference and measured values in the block PD +" Sygives a periodic switching function, which controls achopper Ch that operates as an electronically con-trolled resistance [Slo Fig. 12 shows examples oftorque-speed characteristics.

Motor MI has a squirrel-cage rotor that makes useof the skin effect. The rotor is designed so that a voltage

PO+Sy

Pr Fig. 11. Block diagram of thefrequency-analog control for theconveyor belt. Motor Mi pro-vides the torque, motor M2 de-termines the speed. Ma drivingmotor for the printing cylinderPr. GB gear-box. The blockPD + Sy controls the speed ofthe motor M2 from the signalsarriving via the chopper Cb andalso provides a signal for thecontrol ofthe torque of MI. Thesynchronization circuit andphase-shifter Sy + PhS syn-chronizes the belt with the print-ing cylinder and ensures thecorrect relative angular settingswith the aid of the pulses fromthe proximity switches PSW2for the belt and Psw« for thecylinder. G generator for internalreference frequency.


of mains frequency induced in the stationary cage willonly produce a current in the outer part of the cross-section of the conductor bars, because of the skin effect.This means that the stationary rotor has a relativelylarge effective resistance and hence a high startingtorque. When the motor runs up to speed the frequencyof the rotor current decreases, so that the skin effectalso decreases and the effective rotor resistance be-comes smaller. Because of this interaction the motorhas a constant torque over a wide range of speeds; thisindependenee of speed holds for a considerable rangeof stator-voltage variation, produced for example byphase control (seefig.13).The phase control for the torque motor MI is

driven by a quasi-steady-state signal derived from theswitching function ~ of the speed motor M2, which isproportional to the load (see figs. 11 and 12). Thisensures that while changes in the load of the conveyorbelt are sensed only by the frequency-analog control,they are shared between the two motors in an adjustableratio. The torque distribution is made independent ofthe speed by using a multiplier circuit that takes theslip into account in generating the control signals.

Since the two motors MI and M2 are both mechanicallycoupled by the belt and electrically coupled by the torque-shar-ing, precautions have to be taken to avoid instability. The dynamiccharacteristics can best be discussed with the aid of the controldiagram of fig. 14. The frequency-analog controlof M2 has atransfer function Gsc (to a good approximation the same as for aPI control). The driving torque T2 produced in M2 first of allcompensates the load Tl2 and secondly it operates on the entirespring/mass system (Gl) of the conveyor belt; the angularvelocity cannot therefore be derived directly by simple integrationof the acceleration. At the same time a signal e is derived fromthe control error of the frequency-analog control: this is thephase angle between reference and measured frequencies, and isrepresented by the mark/space ratio of the switching function;this signal e determines the torque of MI. This torque Tl alsooperates both on the braking load Til and on the spring/masssystem (Ga). The speed of each motor shaft is however stillaffected by the torque of the other motor through the springtmass system (G2 = G4).

The transfer functions GI to G4 are determined by the inter-action of the masses and springs in the conveyor belt, seen fromthe appropriate reference system. They are rational fractionalthird-degree functions of the complex frequency.This system can easily be stabilized by putting the torque-

control signal s derived from the frequency-analog controlthrough a lowpass filter of transfer function 1/(1 + ST), toattenuate the higher-frequency components responsible for theinstability. An apparent disadvantage here is that the controlofthe torque in MI is then relatively slow compared with the speedof reaction of the frequency-analog control. However, since thecomplete system can be considered to be a slow system, cuttingoff the frequency response with a lowpass filter does give satis-factory stabilization. If faster operation were required for adifferent kind ofinstallation, effective damping could for examplebe obtained with velocity feedback.

25r/s

Fig. 12. The torque To of the motor M2 (fig. 11) as a function ofthe speed 11; the rotor circuit is closed for a percentage ij of thetime. The motor has four poles and is rated at 800 W.

20Nma=1t.So

125°

15 110°Te

90°tI

70°I55°

10-n

Fig. 13. Torque-speed characteristics of the squirrel-cage motorMI with skin effect, for phase control at various angles IX. Thetorque is practically constant for a wide range of speeds.

+

Fig. 14. Control block diagram of the conveyor-belt drive. Gsccontrol circuit. e error signal. Gml,2 motors MI,2 with theircontrol. GI,Ga driven spring/mass systems of the conveyor belt.Wrl,2 angular velocities of the motors MI,2. Instability arisingfrom the coupling between MI and M2 through the belt can bedamped out either by including a lowpass filter 1/(1 + ST) in thecontrol of M I or by adding a signalof negative sign and propor-tional to the angular frequency Wrl.


Phase adjustment and synchronization at the correctphase angle

For the conveyor belt to make the transition fromthe non-synchronized to the synchronized state, all thatis required is to switch over from the internal referencefrequency to the external frequency derived from thespeed of the printing press. This would introduce asmall transient because of the discontinuity in phase.However, the situation becomes more difficult with

the requirement that both drives must be synchronizedwith a particular phase angle between them, with nomore than about 3 seconds delay between the start ofthe synchronization process and its completion. Acondition for the fulfilment of this requirement is thatthere must be a difference in speed between the printingpress and the dryer. There is a fixed relation betweenthis difference and the time necessary for synchroniza-tion; the shorter this time has to be, the greater thedifference in speed required, but the more marked thetransient at the instant of synchronization. Some sortof monitoring device is also required that measures therelative angular position at the start of synchronizationand compares it with the specified value. If the twovalues do not coincide, which is usually the case, thenthe change in the relative angular difference due to thedifference in speed must be added to the value firstmeasured until it reaches the specified value. Only thencan the internal frequency be exchanged for the exter-nalone.Although solutions to such problems have indeed

been found, whether by using an analog method withservomotors, or a digital method whose pulse ratecorresponds to the required setting accuracy for theangle, neither of these two kinds of sensing device arerequired in frequency-analog drives. Only simple pulsegenerators are required which indicate the separationin time between the successive carriers and between theseparate revolutions of the printing cylinder. The finesubdivisions for accurate indication of the intermediatepositions between two carriers are given by the meas-ured frequency of the motor itself.

Starting from the frequency relation for the inductionmotor:

fr =fm-np,

where fr is the frequency of the rotor currents, used asthe measured frequency.j's, the mains frequency, 11 thenumber of revolutions per second and p the number ofpole pairs, it can be shown that the displacement of thetrain of reference pulses results in a mechanical angularrotation of 3600/pa in one pulse interval T. Here a is afactor by which the measured frequency has to bemultiplied. In the case described here a is equal to 6;this is because the zero crossings that occur in each of

the three rotor phases are used in the multiplicationprocess. For a four-pole machine and a = 6 the pulseinterval therefore corresponds to an angular incrementof 30°.A displacement of the measured signal with respect

to the reference signalof one pulse interval thereforecorresponds to a 30° rotation of the motor shaft. If themotor is connected to gearing, then this corresponds toan angular difference of 300/mg at the output side,where mg is the gear ratio. This increment of 300/mgindicates the fineness of the subdivision achieved bythe frequency-analog measured signals, and hence themaximum setting accuracy. In the present case the gearratio mg is equal to 162,giving a subdivision into unitsof 30°/162 = 0.185°. Such a degree of setting accuracyis by no means necessary, so that frequency dividerscan be used to simplifymonitoring and adjustment.The definite relation between pulses and the angular

positions of the motor shaft forms the basis of thecircuit in which the measurement and adjustment ofthe phase take place. The composition of this circuitwill now be described by examining the operations thattake place in it.After the drying-and-printing unit has been switched

on a gate circuit is opened to pass the first pulse fromthe proximity switch PSW3 for the printing cylinder,which releases an .up-down counter C preceded by afrequency divider (seefig. 15). Two signals now appearat the input to the counter: the reference pulses fromthe rotor of the synchro driven by the printing rollerand of frequency fpr = 6 (fm - npr), and the sixth har-monic of the mains 6fm (in the expression for fpr thequantity I1pr is the speed of the printing roller in revo-lutions per second). When the next pulse arrives fromthe proximity switch PSW2 for the conveyor belt thiscounting process is stopped. The contents of thecounter are now proportional to the number of revolu-

OA +

,I I

d'lFig. IS. Unit for measurement and control ofthe phase differencebetween printing cylinder and conveyor belt (fig. 11 PhS). Cup-down counter. DA digital-to-analog converter. Com com-parator. CU control unit. The phase difference is preset with theaid of the resistor network (right).


tions of the synchro and correspond accurately to theangle described by the printing cylinder during themeasuring time. The proximity switch for the printingpress supplies' a pulse for each revolution of the print-ing cylinder, and the proximity switch for the conveyorbeIt supplies a single pulse each time a carrier passes(to give a phase angle of 200 between the separatecarriers 18 teeth are required on the proximity-switchdisc). The angular position ofthe printing cylinder withrespect to the carriers therefore follows from the angledescribed by the printing cylinder between the twopulses.

Next, the counter input to which the sixth harmonicof the mains frequency is applied is instead switched tothe measured frequency fdr from the drying unit; thisprovides a continuous measurement of the differencein angle between the two machines. This relative anglevaries continuously owing to the difference in speed.When the angular difference reaches 2n (correspondingto 108 pulses) the counter is set back to zero andstarted again because of the periodicity of 2n onrotation.

As soon as the angle - expressed by the state of thecounter - between the cylinder and the belt, rotatingat their different speeds, has become equal to thespecified value, synchronization takes place as follows.The state of the counter is converted into an analogsignal in a digital-to-analog converter (DA, fig. 15) andapplied to the input of an operational amplifier. Cur-rents that represent the specified value of the phaseangle, but of the opposite polarity, are also supplied atthe same input, so that comparison of these and themeasured values is made right at the input of theamplifier. The control error is amplified and applied toa comparator Com that can distinguish between threestates: positive error, negative error and equality(within a tolerance interval). Once the error has becomeequal to zero, the inputs of the counter are disconnectedfrom the pulse generators and the internal referencefrequency of the generator G (fig. 11) is exchanged forthe external one from the printing unit.

In practice it is often necessary to alter the phaserelation between printing and drying units while theyare in operation. However, since the two units arealready synchronized, there is no accelerating ordecelerating reference frequency. This can then bederived from the reference frequency produced by theprinting cylinder or from the measured frequency,with the aid of two special phase-shifters in which theapplied pulses can be delayed as required.

The phase-shifters consist essentially of monostabiecircuits, connected in the path of the reference andmeasured pulses, with variable delay times. By in-creasing these delay times continuously the pulses can

· ·· ·· ·· ·.. 1Ii:ll ·..~ II!!I' 1:11- ·- ·

11:::::1 ·=:~ C::::I

,

Fig. 16. The output voltage of the digital-to-analog converter atsynchronization (left) and at the first phase check (right). Afterthe pulse from the printing-unit proximity switch PSW3 the coun-ter is set back to zero (upper step); next the angular incrementsof the printing cylinder are added up in the counter (short steps)until a pulse arrives from the conveyor-belt proximity switchPSW2. The number of increments indicates the phase angle of thecylinder with respect to the belt. From this moment on the twoangular velocities are compared and only the differences in theangular increments are counted in (long steps). The specifiedphase angle is reached in due course because of the differences inangular velocity. Synchronization takes place here; the counterremains at its final value. At the first check the phase angle ismeasured again; a phase correction of one angular increment isfound necessary to correct a phase error that has arisen duringthe switchover.

be given a steadily increasing delay. This will causethe belt to speed up or slow down, depending onwhether the variable delay is applied to the measured-frequencyor reference-frequency path. The fund amen-tal limit to the delay, and hence the phase shift, that isencountered when the next trigger pulse arrives beforethe previous delayed pulse has been delivered is avoidedby setting the monostabie circuits back to zero andsuppressing the next delayed pulse to arrive.These automatic phase-shifters come into action as

soon as an angular error appears, due for example toa change in the specified phase. After the error hasdisappeared the phase-shifters switch themselves outof use automatically.This system for phase measurement and control also

includes a timer, which compares the value of thephase angle with the prescribed value every ten seconds.For this purpose the existing phase relation is firstdetermined, as in the synchronization process. Anydeviation from the specified relation is corrected byswitching in the phase-shifters.To illustrate the operation of the synchronization

process, the output signal from the digital-to-analogconverter is shown infig. 16. The short 'steps' indicate


the measurement process for determining the angulardifference at the time of the first pulse from the con-veyor belt, the long 'steps' are angular corrections thatare carried out before synchronization takes place.The first periodic phase check can be seen at the endof the trace after the first interval of 10 seconds. Thischeck consists of a measurement and a correction ofone step that had been lost because of the time takenby the synchronization process and now has to bemade up.Provided that the synchronization command does

not coincide with switching on the complete printing-and-drying unit, but is only given when the machinesare running at full speed, the synchronization timeconsists only of the correction time, which depends onthe difference in speed. The delay arising between thesynchronization command and the arrival of the firstpulse from the printing unit and the measurement timethat elapses between the appearance of the printing-unit pulse and that of the conveyor-belt pulse do notextend the synchronization time.To obtain an estimate of the worst-case value of the

synchronization time, we have to remember that thiscan be equal to the reciprocal of the difference in thenumber of revolutions per second of the printingcylinder and the belt. The reason for this is to be foundin the periodicity of a rotation: at the instant when thesynchronization command is applied the phase anglemay have just overshot the specified value, so that an

extra angular difference of nearly 2n must be describedbefore the specified value is reached. This means thatwith a printing-roller speed of70 rev/ruin and a drying-unit speed of 100 rev/min the difference between thetwo is 30 rev/ruin or t tev]« and the maximum syn-chronization time is therefore 2 seconds.The specified phase angle is indicated by a switch

that enables the specified value to be selected with theaid of a resistor network with 16 steps. Since the totalspacing between carriers is 20° this means that thepositioning accuracy is about 1.25°. Inside each of thesteps there is a further uncertainty of ± 0.1°, whichdepends on the sequence in which the first measuredand mains-frequency pulses happen to appear after thestarting pulse for the measurement of the angle.

Summary. The well known phase-lock-loop technique fromelectronics can also be used with advantage in the speed controlof electric motors. In this application thé angular position of themotor shaft is continuously compared with the phase angle of areference signal. The angular differences that arise are used togive a corresponding control of the motor, e.g. by phase controlwith thyristors. The phase comparison allows very accuratetracking to be obtained. The specified value, provided in the formof a frequency, can readily be transferred to a remote locationand the digital data processing is not sensitive to interference.The principle can be applied with both d.c. motors and asyn-chronous motors; with asynchronous motors of the slip-ringarmature type the rotor frequency provides the necessary meas-urement information for speed and phase control.A tracking control for a drying and printing unit for tinplate

sheets is described in detail as a practical example; a specialfeature of this system is that the relative phase can be alteredwhen the unit is in operation.

272 Philips tech. Rev. 33, No. 8/9

Recent scientific publications

These publications are contributed by staff of laboratories and plants which form part ofor co-operate with enterprises of the Philips group of companies, particularly by staff ofthe following research laboratories:

Philips Research Laboratories, Eindhoven, Netherlands EMullard Research Laboratories, RedhilI (Surrey), England MLaboratoires d'Electronique et de Physique Appliquée, 3 avenue Descartes,

94450 Limeil-Brévannes, France LPhilips Forschungslaboratorium Aachen GmbH, WeiI3hausstraJ3e, 51 Aachen,

Germany APhilips Forschungslaboratorium Hamburg GmbH, Vogt-Kölln-StraJ3e 30,

2000 Hamburg 54, Germany HMBLE Laboratoire de Recherches, 2 avenue Van Becelaere, 1170 Brussels

(Boitsfort), Belgium BPhilips Laboratories, 345 Scarborough Road, Briarcliff Manor, N.Y. 10510,

U.S.A. (by contract with the North American Philips Corp.) N

Reprints of most of these publications will be available in the near future. Requests forreprints should be addressed to the respective laboratories (see the code letter) or to PhilipsResearch Laboratories, Eindhoven, Netherlands.

w. Albers: Oriented eutectic crystallization.Preparative methods in solid state chemistry, editorP. Hagenmuller, Academic Press, New York 1972,pp. 367-399. E

W. Albers & L. van Hoof: Aligned eutectic thin filmgrowth.J. Crystal Growth 18, 147-150, 1973 (No. 2). E

G. A. A. Asselman, J. Mulder & R. J. Meijer: A high-performance radiator.Proc. 7th Intersociety Energy Conversion Engng.Conf., San Diego 1972, pp. 865-874. E

H. Bacchi & J. Denis: Electronique associée à deséquipements de photographie ultra-rapide.Acta Electronica 15, 317-327, 1972 (No. 4). L

C. Belouet: About the crystalline perfection of Nd-doped YAG single crystals.J. Crystal Growth 15, 188-194, 1972 (No. 3). L

G. Bergmann: Überlegungen zur Ausnutzung der Emis-sionspolarisation für die polarisierte Kraftfahrzeug-Beleuchtung, I, n.Lichtteehuik 25,21-25 & 62-65, 1973 (Nos. 1 & 2). A

M. Berth & C. Venger: Photodétecteurs rapides à étatsolide.Acta Electronica 15, 281-308, 1972 (No. 4). L

J. W. M. Biesterbos & J. Hornstra: The crystal struc-ture of the high-temperature, low-pressure form ofRh203.J. less-common Met. 30, 121-125, 1973 (No. I). E

R. Bleekrode & W. van Benthem : Resonance fluores-cence of Mg atoms in the gas phase.Philips Res. Repts. 28, 130-132, 1973 (No. 2). E

o. Boser: Influence of high concentrations of soluteatoms on the critical flow stress of binary alloys, I.Theoretical foundations, Il. Application to silver-,gold-, and copper-based alloys.J. app!. Phys. 44,1033-1037 & 1038-1043,1973 (No. 3).

N

H. Bouma (Institute for Perception Research, Eind-hoven): Visual interference in the parafoveal recogni-tion of initial and final letters of words.Vision Res. 13, 767-782, 1973 (No. 4).

J.-P. Boutot: Photomultiplicateur ultra-rapide à galettede microcanaux: le HR 300.Acta Electronica 15, 271-279, 1972 (No. 4). L

M. J. G. Braken: Het fabriceren van schotelbalgele-menten.Lastechniek 39, 4-6,1973 (No. I). E

J. C. Brice & B. A. Joyce: The technology of semi-conductor materials preparation.Radio & Electronic Engr. 43, 21-28,1973 (No. 1/2). M

J. C. Brice & P. A. C. Whiffin: The platinum metals incrystal pulling. Crucibles and other apparatus.Platinum Met. Rev. 17,46-51,1973 (No. 2). M

H. H. Brongersma & P. M. Mul: Analysis of the outer-most atomic layer of a surface by low-energy IOnscattering.Surface Sci. 35, 393-412, 1973. E

E. Bruninx: The scintillation chamber: a high efficiencydetector insensitive to source displacements.Nuc!. Instr. Meth. 106,613-614, 1973 (No. 3). E

T. M. Bruton: The growth of single crystals by thermaldiffusion.J. Crystal Growth 18, 269-272, 1973 (No. 3). M

Philips tech. Rev. 33, No. 8/9 RECENT SCIENTIFIC PUBLICATIONS 273

F. M. A. Carpay & W. A. Cense: Production of in situcomposites by unidirectional crystalline decompositionof non-crystalline solids.Nature phys. Sci. 241, 19-20, 1973 (No. 105). E

G. Clément, C. Delmare (C.E.A., Limeil) & C. Loty:Le Picoscope à microcanaux: un tube à balayage àhaute résolution temporelle pour l'examen de phéno-mènes lumineux dans Ie domaine de la subnanoseconde.Acta Electronica 15, 369-371, 1972 (No. 4). L

P. J. Courtois, F. Heymans & D. L. Parnas (Carnegie-Mellon University, Pittsburgh): Comments on 'A com-parison of two synchronizing concepts by P. B. Hansen' .Acta inform. 1, 375-376, 1972 (No. 4). B

P. A. van Dalen: Electric interaction between trans-verse electroacoustic waves in piezoelectrics of theBleustein-Gulyaev geometry and a semiconductor.Philips Res. Repts. 28, 185-209, 1973 (No. 3). E

M. Davio & J. J. Quisquater: Iterative universallogicalmodules. ,Philips Res. Repts. 28, 265-293, 1973 (No. 3). B

M. Davio & A. Thayse: Representation of fuzzy func-tions.Philips Res. Repts. 28, 93-106, 1973 (No. 2). B

M.-A. Deloron: Photodiode ultra-rapide.Acta Electronica 15, 265-270, 1972 (No. 4). L

J. P. Deschamps: Parametrie solutions of Booleanequations.Discrete Math. 3, 333-342, 1972 (No. 4). B

J. P. Deschamps: Partially symmetric switching func-tions.Philips Res. Repts. 28, 245-264, 1973 (No. 3). B

P. A. Devijver: The Bayesian distance. A new conceptin statistical decision theory.Proc. 1972 IEEE Conf. on Decision and Control and11th Symp. on Adaptive Processes, New Orleans,pp. 543-544. B

J. W. F. Dorleijn, W. F. Druyvesteyn, G. Bartels &W. Tolksdorf: Magnetic bubbles and stripe domainssubjected to in-plane fields, I. Uniaxial anisotropy,11. Contribution of the cubic anisotropy.Philips Res. Repts. 28, 133-151 & 152-157, 1973(No.2). E, H

J. Durieu & Y. Genin: On the existence of linear multi-step formulas enjoying an 'h2-process' property.Philips Res. Repts. 28, 120-129, 1973 (No. 2). B

D. den Engelsen: Transmission ellipsometry and po-larization speetrometry of thin layers.J. phys. Chem. 76, 3390-3397, 1972 (No. 23). E

G. Engelsma: A possible role of divalent mangeneseions in the photoinduction of phenylalanine ammonia-lyase.Plant Physiol. 50, 599-602, 1972 (No. 5). E

G. Eschard: Tube obturateur de type diode pour photo-graphie ultra-rapide.Acta Electronica 15, 309-316, 1972 (No. 4). L

L. J. M. Esser: Peristaltic charge-coupled device: anew type of charge-transfer device.Electronics Letters 8,620-621, 1972 (No. 25). E

W. G. Essers, G. Jelmorini & G. W. Tichelaar: Elec-trode phenomena with plasma-MIG welding.Proc. 2nd Int. Conf. on Gas Discharges, London 1972,pp. 135-137. E

E. Fabre: Bulk effects in photoluminescence meas-urements.J. appl. Phys. 43, 3788-3789, 1972 (No. 9). L

D. G. J. Fanshawe: Sound field plotter.SERT J. 7, 37-38, 1973 (No. 2). M

A. J. Fox: Longitudinal electro-optic effects in bariumstrontium niobate (BaxSrl-xNb206).J. appl. Phys. 44, 254-262, 1973 (No. I). M

G. Frens: Controlled nucleation for the regulation ofthe particle size in monodisperse gold suspensions.Nature phys, Sci. 241, 20-22, 1973 (No. 105). E

C. J. Gerritsma & P. van Zanten: The dependence ofthe electric-field-induced cholesteric-nematic transitionon the dielectric anisotropy.Physics Letters 42A, 127-128, 1972 (No. 2). E

C. J. Gerritsma & P. van Zanten: An explanation oftheobserved field-induced blue shift of an imperfectlyaligned planar cholesteric texture.Physics Letters 42A, 329-330, 1972 (No. 4). E

J. Graf: L'intensificateur d'images à microcanaux.Applications à la photographie ultra-rapide.Acta Electronica 15, 357-362, 1972 (No. 4). L

P. Guétin & G. Schréder: Tunneling spectroscopy andband-structure effects in n GaSb under pressure.Phys. Rev. B 6, 3816-3835, 1972 (No. 10). L

F. E. L. ten Haaf (Philips Nuclear Applications La-boratory, Eindhoven): Colour quenching in liquidscintillation coincidence counters.Liquid Scintillation Counting 2, 39-48, 1972.

J. Hallais, C. Schemali & E. Fabre: Vapour epitaxialgrowth and characterization of InAsI-XPX'J. Crystal Growth 17, 173-182, 1972. L

H. Haug & K. Weiss: Influence of a thin surface-dislocation layer on the Kapitza resistance.Proc. 4th Int. Cryogenic Engng. Conf., Eindhoven1972, pp. 129-130. E

J. C. M. Henning: Exchange measurements by meansof electron spin resonance and optical techniques.Pulsed magnetic and optical resonance, Proc. AmpèreInt. Summer School n, Basko polje 1971, pp. 179-205;1972. E

D. Hennings & G. Rosenstein: X-ray structure inves-tigation of lanthanum modified lead titanate with A-site and B-site vacancies.Mat. Res. Bull. 7, 1505-1513, 1972 (No. 12). A

K. Herff & E. Roeder : The investigation and meas-urement of thin layers.Pract. Metallogr. 9, 615-623, 1972 (No. 11). A

274 RECENT SCIENTIFIC PUBLICATIONS Philips tech. Rev. 33, No. 8/9

w. A. H. J. Hermans (Philips Medical Systems Divi-sion, Eindhoven): On the instability of a translatinggas bubble under the influence of a pressure step.Thesis, Eindhoven 1973. (Philips Res. Repts. Suppl.1973, No. 3.)

J. H. C. van Heuven & T. H. A. M. Vlek: Anisotropyin alumina substrates for microstrip circuits.IEEE Trans. MTT-20, 775-777, 1972 (No. 11). E

B. Hoekstra & R. P. van Stapele: Anomalous magneticanisotropy and resonance linewidth in CdCr2S4.Phys. Stat. sol. (b) 55, 607-613, 1973 (No. 2). E

J. P. Hulin: Parametrie study of the optical storageeffect in mixed liquid-crystal systems.Appl. Phys. Letters 21, 455-457, 1972 (No. 10). L

B. A. Joyce: Some aspects of the surface behaviour ofsilicon.Surface Sci. 35, 1-7, 1973. M

B. A. Joyce & J. H. Neave: Electron beam-adsorbateinteractions on silicon surfaces.Surface Sci. 34, 401-419, 1973 (No. 2). M

F. Kettel: The oxidation of zirconium at high tem-peratures.Philips Res. Repts. 28, 219-244, 1973 (No. 3). A

E. Klotz & M. Kock: A simple ground-glass correlator.Optics Comm. 6, 391-393, 1972 (No. 4). HM. Kock & U. Tiemens : Tomosynthesis: a holographicmethod for variable depth display.Optics Comm. 7, 260-265, 1973 (No. 3). H

A. J. R. de Kock: Microdefects in dislocation-free sili-con crystals.Thesis, Nijmegen 1973. (Philips Res. Repts. Suppl.1973, No. 1.) E

L. J. Koppens: Improved ferrite memory cores obtainedby a new preparation technique.IEEE Trans. MAG-8, 303-305, 1972 (No. 3). E

E. Krätzig: Ultrasonic-surface-wave attenuation ofgapless superconductors.Phys. Rev. B 7, 119-129, 1973 (No. I). H

J.-P. Krumme & P. Hansen: A new type of magneticdomain wall in nearly compensated Ga-substitutedYIG.Appl. Phys. Letters 22,312-314, 1973 (No. 7). HF. A. Kuijpers: Thermodynamic and magnetic proper-ties of RC05 hydrides.Ber. Bunsen-Ges. phys. Chemie 76, 1220-1223, 1972(No. 12). EF. A. Kuijpers: RC05-H and related systems.Thesis, Delft 1973. (Philips Res. Repts. Suppl. 1973,No.2.) EP. K. Larsen & S. Wittekoek: Photoconductivity andluminescence caused by band-band and by Cr3+crystal field absorptions in CdCr2S4.Phys. Rev. Letters 29, 1597-1599, 1972 (No. 24). EJ. H. J. Lorteye: Alfanumerieke displays met gasont-ladingen, lichtgevende diodes en vloeibare kristallen.Ingenieur 84, ET 145-150, 1972 (No. 50). E

B. J. de Maagt & J. F. Verwey: Determination of thecapture cross-section of hole-trapping centres in Si02:Philips Res. Repts. 28, 210-218, 1973 (No. 3). E

H. H. van Mal & A. Mijnheer: Hydrogen refrigeratorfor the 20 K region with a La'Nis-hydride thermalabsorption compressor for hydrogen.Proc. 4th Int. Cryogenic Engng. Conf., Eindhoven1972, pp. 122-125. E

A. P. J. Michels: C.V.S. test simulation of a 128-kWStirling passenger car engine.Proc. 7th Intersociety Energy Conversion Engng.Conf., San Diego 1972, pp. 875-886. E

A. Mijnheer: Experiments on a two-stage Stirling cryo-generator with unbalanced regenerators.Proc. 4th Int. Cryogenic Engng. Conf., Eindhoven1972, pp. 83-86. E

A. Mircea, A. Roussel & A. Mitonneau: l/fnoise: stilla surface effect.Physics Letters 41A, 345-346, 1972 (No. 4). L

B. J. Mulder: Preparation of thin single crystals ofchalcosite and other cuprous sulphides.Mat. Res. Bull. 7, 1535-1542, 1972 (No. 12). E

B. J. Mulder: Optical properties of an unusual form ofthin chalcosite (CU2S) crystals.Phys. Stat. sol. (a) 15, 409-413, 1973 (No. 2). E

F. L. van Nes (Institute for Perception Research, Eind-hoven): Motor programming and errors in hand-writing.Neurophysiology studied in man, Proc. Syrnp. Paris1971 (Int. Congress Series No. 253), pp. 420-425; 1973.

G. F. Neumark: Auger theory at defects - applicationto states with two bound particles in GaP.Phys. Rev. B 7, 3802-3810, 1973 (No. 8). N

H. W. Newkirk: Observations on dislocations in tetra-phenyltin and its isomorphs.J. Chem. Soc. Dalton Trans. 1973, 12-14 (No. I). A

S. G. Nooteboom (Institute for Perception Research,Eindhoven): The perceptual reality of some prosodicdurations.J. Phonetics 1, 25-45, 1973.

A. van Oostrom: The influence of temperature and fieldstrength on the evaporation, desorption and emissionfrom solid surfaces.Ned. T. Vacuümtechniek 10, 71-77, 1972 (No. 5). E

G. den Ouden: Internal friction and ductility in theheat-affected zone.Metal Constr. Brit. Welding J. 4, 94-95, 1972 (No. 3).

EL. J. van der Pauw: Impedance matrix of coupledpiezoelectric resonators.Philips Res. Repts. 28, 158-178, 1973 (No. 2). E

R. J. Pedroso & G. A. Domoto (Columbia University,New York): Perturbation solutions for sphericalsolidification of saturated liquids.Trans. AS ME C (J. Heat Transfer) 95, 42-46, 1973(No. I). . N

Philips tech. Rev. 33, No. 8/9 RECENT SCIENTIFIC PUBLICATIONS 275

G. Piétri: High-speed photoelectronics.Acta Electronica 15, .263-264, 1972 (No. 4). (Also inFrench, pp. 261-262.) L

J. Pol man & J. E. van der Werf: Cd(53Pl) and Cd(51PI)atom density decay in a Cd-Ne afterglow plasma.Physics Letters 42A, 153-154, 1972 (No. 2). E

H. Rau, T. R. N. Kutty & J. R. F. Guedes de Carvalho:High temperature saturated vapour pressure of sulphurand the estimation of its critical quantities.J. chem. Thermodyn. 5, 291-302, 1973 (No. 2). A

F. Rondelez: Conductance measurements above asmectic C ++ nematic transition.Solid State Comm. 11, 1675-1678, 1972 (No. 12). L

P. Röschmann: Compact YIG bandpass filter withfinite-pole frequencies for applications in microwaveintegrated circuits.IEEE Trans. MTT-21, 52-57, 1973 (No. I). H

J. A. J.Roufs (Institute for Perception Research, Eind-hoven): Dynamic properties of vision, Ill. Twinflashes, single flashes and flicker fusion.Vision Res. 13, 309-323, 1973 (No. 2).

T. E. Rozzi & G. de Vrij: A series transformation fordiaphragm-type discontinuities in waveguide.IEEE Trans. MTT-20, 770-771, 1972 (No. 11). E

U. J. Schmidt, E. Schröder (Elektro Spezial, Hamburg)& W. Thust (Elektro Spezial, Hamburg): Optimizationprocedures for digital light beam deflectors.Appl. Optics 12, 460-466, 1973 (No. 3). H

A. A. Schüeli (Eidg. Techn. Hochschule, Zürich):Schnelle Parallel-Dateniibertragung mit zeitbegrenztenImpulsen.Thesis, Zürich 1973. (Philips Res. Repts. Suppl. 1973,No.4.)

M. H. Seavey: Observation of light-induced anisotropyin ferric borate by acoustic resonance.Solid State Comm. 12,49-52, 1973 (No. I). E

G. Simpson: Acoustic surface waves on paratellurite.Electronics Letters 9, 21-22, 1973 (No. 2). M

A. Slob, H. A. van Essen & C. M. Hart: Aspecten vanzakrekenmachines.Informatie 15, 1-16, 1973 (No. I). E

J. L. Sommerdijk : Influence of host lattice on the infra-red-excited visible luminescence in Yb3+,Er3+-dopedfluorides.J. Luminescence 6, 61"67, 1973 (No. I). E

D. B. Spencer: A comparison of FM and digital modu-lation for direct television broadcasting from geo-stationary satellites.'B.B.U. Rev. tech. Part No. 137,23-30, 1973. M

W. T. Stacy & U. Enz: The characterization of mag-netic bubble-domain materials with X-ray topography.IEEE Trans. MAG-8, 268-272, 1972 (No. 3). E

S. Strijbos: Burning-out of a carbonaceous residuefrom a porous body,Chem. Engng. Sci. 28, 205-213, 1973 (No. I). E

F. L. H. M. Stumpers, N. van Hurck& J. O. Voorman:Ein Empfänger für Zweiseitenband-AM und für Ein-seitenband-AM mit teilweise unterdrücktem Träger.Rundfunktechn. Mitt. 16, 202-206, 1972 (No. 5). E

D. G. Taylor & P. Schagen: The application of channelimage intensifiers to low light-level television.Adv. in Electronics & Electron Phys. 33B, 945-959,1972. M

A. Thayse: On some iteration properties of Booleanfunctions.Philips Res. Repts. 28, 107-119, 1973 (No. 2). B

J. B. Theeten, J. L. Domange (E.N.S.C.P., Paris)& J. P. HnrauIt: LEED at 20 OK - a compulsory testof theory.Solid State Comm. 11, 1477-1479, 1972 (No. 10). L

W. Tolksdorf: Herstellung hexagonaler Ferrit-Ein-kristalIe mit Y-Struktur aus schmelzfl.üssiger Lösung.J. Crystal Growth 18, 57-60, 1973 (No. I). H

W. Tolksdorf, G. Bartels, G. P. Espinosa, P. Holst,D. Mateika & F. Welz: Controlled lattice constantmismatch by compositional changes in liquid phaseepitaxially grown single crystal films of rare earthyttrium iron gallium garnets on gadolinium galliumgarnet substrates.J. Crystal Growth 17, 322-328, 1972. H

M. J. Underhill: The delay stabilised variable oscillator.A new type of stable tunable oscillator.SERT J. 7, 38-39, 1973 (No. 2). M

H. Vantilborgh & A. van Lamsweerde: On an extensionof Dijkstra's semaphore primitives.Inform. Proc. Letters 1, 181-186, 1972 (No. 5). B

A. A. van der Veeke: High voltage pulse amplifierdrives capacitive loads with short rise times.Rev. sci. Instr. 43, 1702-1703, 1972 (No. 11). EJ. F. Verwey: Emitter avalanche currents in gatedtransistors.Proc. 9th Annual Reliability Physics Symp., Las Vegas1971, pp. 16-24. EJ. Vlietstra & J. B. M. Lucassen (Philips Centre forTechnology, Eindhoven): AEDBAR - a softwaresystem for designing and analysing planar bar mech-anisms.Software Pract. Exper. 3, 29-42, 1973 (No. I).

H. O. Voorman &A. Biesheuvel : An electronic gyrator.IEEE J. SC-7, 469-474, 1972 (No. 6). ,EJ. J. Vrakking & F. Meyer: Quantitative aspects ofAuger electron spectroscopy; on the importance ofattenuation of the primary electron beam.Surface Sci. 35, 34-41, 1973. EK. WaIther: Sound velocity and magnetoelastic coup-ling in CsMnF3.Physics Letters 42A, 315-316, 1972 (No. 4). EiH. Weinerth: Integrated wideband amplifiers forVHF/UHF community antenna applications.Electronic product news, Proc. EPN Seminar 'Prac-tical design with integrated circuits', Paris 1972, pp.135-171; 1973. E

276 RECENT SCIENTIFIC PUBLICATIONS . Philips tech. Rev. 33, No. 8/9

H. W. Werner: Instrumental aspects of secondary ionmass speetrometry and secondary ion imaging massspectrometry.Vacuum 22,613-617,1972 (No. 11). EH. W. Werner & H. A. M. de Grefte: Investigation ofsurface layers by SIMS and SIIMS.Surface Sci. 35, 458-472, 1973. E

G. Wesselink, D. de Mooy & M. J. C. van Gemert:Temperature determination of high-pressure opticallythick gas discharges by a modified Bartels' method.J. Physics D 6, L 27-30, 1973 (No. 4). E

P. W. Whipps, R. S. Cosier & K. L. Bye: Orthorhombicdiglycine sulphate.J. Mat. Sci. 7, 1476-1477, 1972 (No. 12). M

Contents ofPhilips Telecommunication Review 31, No. 2, 1973 (P RX telephone exchange issue):

F. J. Schramel: The evolution of telephony (pp. 46-49).A. W. van 't Slot &M. J. Laarakker: A general description ofthe PRX system (pp. 50-60).J. H. Hiemstra & H. L. Rescoe: System software (pp. 61-72).G. J. Kamerbeek: Call handling procedure (pp. 73-76).J. Borcherding & B. Yff: Traffic considerations and modular system growth (pp. 77-80).W. A. van Dam, H. Schreur & A. C. Steenhuisen: The operational man/machine conversation (pp. 81-84).J. A. Brakel, W. G. Ekas & H. J. Goebertus: Reliability and maintenance aspects (pp. 85-92).A. Timmer: Mechanical design (pp. 93-98).J. de Boer: Traffic studies in the switching network of PRX (pp. 99-105).M. Koeman: A specification method for line equipment in automatic telephony (pp. 106-110).

Contents of Philips Telecommunication Review 31, No. 3, 1973:

W. Milort: Time division switching in telephony (pp. 113-116).R. B. Buchner & W. Milort: The switching network of Philips' PDX system (pp. 118-130).G. van Dasier, H. van Lambalgen & P. M. van Daal: A threshold extension demodulator (pp. 131-146).J. de Boer: Comparison of random selection and selection with fixed starting position in a multi-stage link net-work (pp. 148-155).A. van Dedem: 8 TR 400: a new generation of channel equipment (pp. 156-157).

Contents ofElectronic Applications Bulletin 32, No. 1, 1973:

J. Kaashoek: Deflection system design for 110° shadow-mask tubes (pp. 3-22).o. C. Voss: Development of wideband triodes for band IV/V television transposers (pp. 23-27).G. Wolf: Mains isolating switch-mode power supply (pp. 28-43).A. H. Hilbers: Design of high-frequency wideband power transformers (pp. 44-48).

Contents of Mullard Technical Communications 12, No. 119, 1973:

J. S. Malcolm: F.M. radio front-end with improved large signal handling performance (pp. 262-270).L. E. Jansson: A survey of converter circuits for switched-mode power supplies (pp. 271-278).B. George: Variable 35V lOA switched-mode voltage regulator (pp. 279-292).

Contents of Mullard Technical Communications 12, No. 120, 1973:

L. E. Jansson: Radio frequency interference suppression in switched-mode power supplies (pp. 294-298).J. M. Lavallee: D.C. motor reversible drive using regenerative braking (pp. 299-339).

Volume 33, 1973, No. 8/9 pages 213-276 Published 25th January 1974

frequency-analog speed control - philips bound... · frequency-analog speed control ... used...

Documents