258 PHILlPS TECHNICAL REVIEW VOL. 17, No. 9

STEREO REVERBERATION

by R. VERMEULEN. 534.844: 534.846.4: 621.395.625.3

While too long a reverberation time makes speech unintelligible, a reverberation time which istoo'short makes music sound "dry" and brittle . .JI,lany varieties of acoustic materials areavailable for shortening the reverberation time and improving the intelligibility. Lengtheninga/the reverberation time and --:- perhaps more important - making the sound diffuse, canbe achieved by electro-acoustic means. Tèsts have shown that in this way a good theatre' hallcan be made suitable for concerts.

Introduction .

The acoustic properties of a hall are determinedby the behaviour of the sound waves, in particularthose which are reflected from the walls. The im-portance" of the reflected sound can be well under-stood if one has listened to an open-air speech madewithout an amplifier installation, Ol'if one imaginesan open-ail' performance by a symphony orchestrawithout this aid.

A simple calculation will show that in a hall ateven quite a small distance from the orchestra, thedirect sound can sometimes be weaker than thereflected sound. At a distance T from a soundsource of power P, the energy density of the directsound is equal to P/(4nT2e), where cis tbc velocityof sound. In a hall with a volume V and reverber-ation time T (defined as the time in which the soundintensity decreases by 60 dB, after the source hasceased radiating), the energy density of the indirectsound 1) is PT/(13.8 V). These two energy den-sities are equal at a distance TO = (1.1 V/cT)!.Vahies of TO corresponding to various practicalvalues of Vand T are as follows:

V= 100T= 0.7

10001.0

10000m3

1.5 s

TO = 0.7 1.8 4.5m.

At distances greater than TO the indirect soundpredominates. It is seen that this can occur atdistances of only a few metres.The qualities required of a hall for speech and

for music are quite different. In a theatre, the intel-ligibility is of primary importance. If speech is tobe clearly understood," the reflected sound mustreach the audience with so little delay that it rein-forces the' direct sound but does not overlap the

1) See for example A. Th. van Urk, Auditorium 'acousticsand reverberation, Philips tech. Rev. 3, 65-73, 1938.

sounds which follow. For the latter, the persisteneeof the preceeding _sound must be regarded as"background noise", which adversely affects intel-ligibility. As a rough guide, one can say that allsound which reaches the audience within 50 milli-seconds can be regarded as useful sound 2). ErwinMeyer has formulated the idea of "clearness" or"definition" 3), which he defines as follows:

50 msec

J p2(t) dto

where p is the sound pressure, t is the time (mea-sured from the moment at which the source issilenced) and tI is a time much g~eater than 50msec.

For a concert hall, on the other hand, the firstrequirement is not intelligibility, but a fine, fulltone. Here it is much more difficult to specify therequirements. For speech, the reverberation must beaccepted as an inevitable, disturbing accompani-ment to the useful sound, because it simply is notpossible to silence the sound suddenly after 50 msec.For music we know that the reverberation timenot only may he, but must be, longer. The optimumvalue is clearly dependent on the nature of themusic. It is often seen that a composer has cons-ciously taken into account the acoustics of the space(church, concert hall, room) where he wanted hismusic to be played. The inclination to sing in thebathroom can probably be largely attributed to thelong reverberation time of this acoustically "hard"room.

It is becoming increasingly clear that the rever-beration time is not the only property governingthe suitability of a hall for' musical performances.

2) See the article referred to in 1), pp. 72-73. '3) E. Meyer, Definition and diffusion in rooms, J. Acoust.

Soc. Amer. 26, 634, Sept. 1954.

.J'[ARCH 1956 STEREO REVERBERATION 259

One might even conjecture that here, too, reverber-ation is merely an inevitable subsidiary effect.Just as important, or perhaps even more so, is the"diffuse-ness" of the sound (and possibly also thenature of the fluctuations of the reverberation).

To study these phenomena more precisely, weattempted to produce an artificial diffuse reverber-ation in the laboratory by means of distributedloudspeakers which repeated the music played,with controllable intensity and lag. This experi-ment appeared to improve the acoustics to such anextent that we ventured to take the bold step ofusing this artificial reverberation to make a theatresuitable for concerts. We propose that this artificialdiffuse reverberation be called "stereo reverbe-ration".

Our first installation was in the Philips Theatreat Eindhoven, whose acoustical properties as atheatre were very satisfactory as a result of rebuil-ding in 1935, but which left much to be desiredas a concert hall. In addition to this theatre, a hallknown as the "Gebouw voor Kunsten en Weten-schappen" (Arts and Sciences Hall) in The Hague,is now fitted with a permanent installation for stereoreverberation 4).

Principle of the installation for stereo reverberation

The delay wheel

The principle of the stereo reverberation instal-lation may be explained with the help of fig. 1.Controllable time lags are obtained by means ofmagnetic recording and playback. Magnetic materialsuch as is used for magnetic tape is coated on therim of a wheel ("delay wheel") 5). The music isrecorded on this material via a microphone and arecording head. A number of play-back heads- in the final apparatus four, in experimentaltypes six (fig.2) or more (fig.3) - are mountedaround the circumference of the wheel and con-nected Vla separate channels to loudspeakers.These are installed in various places in the hall:

4) Demonstrations of stereo reverberation have also beengiven at the first LC.A. Congress on Electro-acoustics,(Acustica 4, 301, 1954), at Gravesano in Switzerland (atthe invitation of the conductor W. Scherchen; see his bookMusik, Raumgestaltungund Elektroakustik, Arsviva Verlag,Mainz 1955) and at the 3rd "Tonmeistertagung" of theNordwestdeutsche Musikakademie Detmold (see D. Kleis,Elektron. Rdsch. 9, 64-68, 1955 (No. 2». Similar tests,but done in the open air, are described by H. S. Knowles,Acustica 4, 80-82, 1954 (No. I).

5) Others have also constructed a similar equipment, but havenot created diffuse reverberation with it; see H. Schiesser,Einrichtungen znr Erzeugnng künstlichen N achhalls,Funk und Ton 8,361-368;1954 (No. 7), and P. Axon andeo-workers, Artificial reverberation, J. Instn. El. Engrs. 1,368-371,1955 (No. 6).

on the ceiling, along the balustrade of the balcony,in a lighting cornice, in "dead" corners under thebalcony, etc. Between the last play-back head and

Fig. 1. Installation for simulating indirect sound with varioustime lags. A auditorium. P platform. AI microphone. Wdelay wheel, coated on the edge with magnetic materialsuitable for magnetic sound recording. 0 recording head,1... 6 play-back heads. 7 erasing head. L loudspeakers.

Fig. 2. Stereo reverberation installation in the Philips Theatreat Eindhoven. At the bottom of the cabinet is the delay wheel(cf. fig. I); above it the amplifiers.

260 PHILIPS TECHNICAL REVIEW VOL. 17, No. 9

the recording head is au erasing head, which en-sures tbat the magnetic layer is blank on returningto the recording head.

For three reasons, however, the listener hearsmore than these seven (1 + 6) reports: firstlybecause to each play-back bead, several loud-

Fig. 3. Stereo r everberat ion installation in the "Cebou w voor Kunsten en Wetenschappen"(Arts and Sciences Hall), The Hague, in experimental form. The delay wheel was thenfitted with ten play-back heads, of which six could operate sirnultaneousl y. It was possibleto switch over rapidly from one set of play-back heads to another.

Let us consider the case of a sharp report pro-duced in the hall (pulse 10, fig. 4). After a certaintransit time this reaches the microphone and thenthe artificial indirect sound begins. The latter, if weare using six play-back heads, consists in tbe firstplace of six successive reports from the loudspeakers(11 •.. 16), The intensity of each of these reportscan be adjusted at will by the gain controls; thetime intervals are also under control, by thespacing of the heads round the wheel and by itsspeed of revolution.

Fig. 4. If a report (pulse Jo) is produced in a hall with stereoreverberation, tbc six loudspeakers deliver during one revo-lution of the delay wheel the six reports 11.. .16, If electricalfeedback is applied from the sixth play-back head to the re-oording head, the loudspeakers give a second seri.es of sixreports (11' ... 16'), a third, and so on. (Acoustic feedback isbere neglected.)

speakers are connected, which are dispersed through-out the hall and are at different distances from thelistener, so. that the transit times are different;secondly because the walls reflect both the originalsound and those coming from the loudspeakers,and thirdly because the report from each loud-speaker is picked up by the microphone, oncemore recorded on the wheel and, tbough attenu-ated, reproduced six-fold.All these effects contribute to the fact that the

original report is followed by a large number ofothers, so that the reverberation time can achieveconsiderable values. The third effect mentionedabove - the feedback from the loudspeakers to themicrophone - even involves the danger that oneparticular note continues to sound too long andin the extreme case howls back continuously;therefore this effect must be curtailed as much aspossible (we shall return to this point presen tly).This acoustic feedback can be advantageouslyreplaced by electrical feedback, in which a cbosenfraction of the signal from the last play-back headis fed back to the recording head. This records thesignal afresh: in our example, a second series ofsix reports then ensues (11' ... 16' in fig. 4); this seriesis once more recorded and a third series follows, and

)lARCH 1956 STEREO REVERBERATION 261

so on. By adjusting the fraction of the outputsignal which is fed back, each series can be atte-nuated to any given degree, so that the reverbe-ration time can be chosen at will.

Calculation of the effect of the stereo reverberation onthe acoustic properties of the hall

The following calculation shows, for a simplecase, the effect which stereo reverberation has on theacoustic properties of the hall. Suppose that theenergy density in the hall is E(t). Then- VdE/dt isthe rate at which acoustic energy in the hall dimin-ishes. In a. hall without stereo reverberation, thismust be equal to the power absorbed by thewalls 6), which is proportional to E, say equal toaE. Stereo reverberation supplies extra powerproportional to the energy density at some time •ago: f3E(t-.). We can thus constructtheequation:

dE .- V- = aE(t) - f3E(t,_.).

dt

A solution to this is:

E = Eo exp(-mt),. where

mV = a - f3 exp(ni.) ;

the quantitity m is inversely proportional to thereverberation time T, and equal to 13.BIT. For smallvalues of mx, an approximate value of exp(mr]is given by I + m•. Thus we have:

a- f3

V+f3.I

From this it is seen that increasing the strength ofthe stereo reverberation (f3) has the same effectas decreasing the absorption (a) or increasing thevolume (V) of the hall. An increased lag. also givesthe effect of a hall of larger volume; we shall returnlater to this point.

The microphones

. The above calculation was based on the assump-tion that one could speak of "the" energy densityin the hall. In many cases the total energy (sum ofpotential and kinetic energy) is indeed fairly evenlydistributed throughout the hall. With standingwaves, however, as is well known, the p<?tential

6) See for example the article referred to in 1), where it isshown that the proportionality factor a is equal to 1/4UAc,where a is the 'average absorption coefficient and A thearea of the walls.

and kinetic energy alternate with each other, andthis means that the sound pressure at constantfrequency changes sharply from place to place,and, conversely, that at a partienlar place thesound pressme is very dependent on the frequency.The microphone which picks up the signal whichis fed back to the hall via the delay wheel andthe loudspeakers, responds only to the pressureat the spot. The factor f3 in our calculation there-fore varies sharply with the frequency, and thereverberation time, which is inversely proportionalto a- f3, varies also and to a much greater degree.If f3 as a function of the frequency shows a peak,then increasing the amplification will cause thenote at which the peak occurs to go on soundingfor a long time, while for most other frequencies,the reverberation time is not yet lengthened appre-ciably. With even greater amplification, the notefails to decay at all (howl-back: a- f3 has hecomenegative).The obvious solution is to try and suppress the

peak by a filter in the microphone channel. However,there are so many peaks in the frequency charac-teristic of a hall, and these peaks are so sharp, thatthe suppression of all of them would be a hopelesstask, particularly since the peaks are modifiedby all changes made in the hall or on the platform. ,It is worthwhile, nevertheless, to attenuate thosefrequency ranges in 'which the highest peaks occurin such a way that increase of amplification causesa number of widely separated frequencies to howlback at the same time.

In this connection it can be argued that electrical feedbackvia the delay wheel is a bctter means of obtaining a longreverberation time than acoustic feedback via the hall.In a system with delayed feedback, the feedback signal and

the input signal exhibit, in general, a relative phase differencewhich is proportional to the product of frequency and timelag. If there are a great number of feedback channels (as thereinvariably are in practice, with acoustical feedback), andif we suppose that they all make equal contributions tothe input signal, then in general the contributions willshow fairly random phase differences, so that the averageresultant increases as the root of the number of feedbackchannels. There will be, however, one or more frequencies forwhich all the contributions are nearly. in phase with eachother. For these frequencies the resultant will be a maximumand equal to the algebraic sum of the contributions, and thusproportional to the number of feedback channels. This reaso-ning makes it clear that the ratio of the maximum value of theresultant (which occurs at a particular frequency and limitsthe maximum amplification) to the average value (at otherfrequencies) increaseswith the root of the number of feedbackchannels. In the case of a microphone and loudspeakers in thesame hall, this number is very large and it is to be expectedthat a situation can easily arise in which one note continuesto"sound for a long time. With feedback via the delay wheel,

262 PHILlPS TECHNICAL REVIEW VOL. 17, No. 9

we are using only one feedback channel (from the last play-back head to the recording head) and the above-mentioneddanger is thus much less. It would be possible to introducemore feedback channels, e.g. from one or more of the precee-ding playback heads to the recording head. Experimentshave shown that this is undesirable, and this is partly ex-plained by the above considerations.

Fig. 5. The stage in the "Gebouw voor Kunsten en \Veten-schappen" in The Hague. MI is the line microphone (rod withten condenser microphones). (Other microphones visible inthe photograph were for broadcasting and had nothing todo with stereo reverberation.)

Instead of a microphone which responds to theaverage sound density in the hall, the other extremewould be one which responds exclusively to soundcoming directly from the source and is insensitiveto sound from the hall 7). The danger of noteshowling back would then he completely averted.This situation can be closely enough approximatedto by using a microphone having a sharp directionalcharacteristic. An arrangement for achieving this,a so-called "line microphone", consists of a groupof ten condenser microphones mounted at equalintervals along a rod rather more than a metrein length (fig. 5). By suitably positioning this line

7) This occurs, for example, in the case of music reproducedin a hall, when it is (or was) actually performed elsewhere.~r e return to this point again at the conclusion.

microphone above the orchestra, we can ensurethat the loudspeaker signal at the microphone,although still making an important contribution,no longer dominates. If it is not possible to coverthe orchestra adequately with one line microphone,two may be used (fig. 6).

The use of one or more extra microphones canbe desirable in order to strengthen weak instrumentsin the orchestra. Thus, for example, in the PhilipsTheatre at Eindhoven the organ was rather weak incomparison with the orchestra and choir in theannual performance of Bach's St Matthew Passion;it appeared to be an improvement when this in-strument was boosted by a microphone of its own(fig. 7). With such cases in mind, the stereo rever-beration system was provided with several micro-phone channels with mixing facilities. These should,however, be used with the utmost restraint, forthe sound engineer must never encroach uponthe conductors prerogative for the balance of theinstruments in the orchestra.

Fig. 6. In the Philips Theatre at Eindhoven two line micro-phones are used (MI' lW2). The loudspeakers are mounted in aconcealed position behind the lighting cornice L (comparefig. 9).

:NIARCH 1956 STEREO REVERBERATION

Fig. 7. The small organ in the Philips Theatre can be reinfor-ced through an amplifier channel fed from a separate micro-phone M3.

The loudspeakers

As already suggested, the diffuseness of the soundis perhaps even more important than the lengthen-ing of the reverberation time. Diffuseness can beobtained by dispersing the loudspeakers over thehall (fig. 8) and connecting them to the variousplay-back heads. The wiring is simplified if all theloudspeakers belonging to one group (fed from thesame play-back head) are connected in series. Inthe case of four play-back heads, one four-corecable is run around the hall, balconies, etc; where aloudspeaker is installed, the appropriate core is cutand the speaker connected in series (fig. 9).The distribution of the loudspeakers between the

various play-back heads should be done as ran-domly as possible. The only restrietion is that theaudience should never get the impression that thesound comes from the loudspeakers. We shall nowtry to explain further the general lines to follow inorder to avoid this impression.We shall use some results of the work of K. de

Boer on stereophony 8). In fact, we are here dealing

8) For a recapitulation of the pr inciples of stereophony, withreferences to the Iiterature, see R. Vermeulen , Philipstech. Rev. 17, 171-177, 1955/56 (No. 6).

Fig. 8. In the "Gebouw voor Kunsten en \Vetenschappen" the loudspeakers are mountedalong the edge of the upper balcony and (not visible in the photograph) under the balconies.

263

264 PI-IILIPS TECHNICAL REVIEW VOL. 17, No. 9

with an analogous problem. The condition thatnobody in the hall may consciously hear musiccoming fi·om the loudspeakers, means that at allpoints in the hall the "sound image" must appearto be located in the orchestra. This latter can be

Fig. 9. The loudspeakers of the stereo reverberation installa-tion in the Philips Theatre are mounted in groups of two orthree in the panels which cover the lighting cornice (L infig. 6). At lone of the panels has been lifted up. 2 is the (six-core ) cable, 3 a junction box.

regarded as onc of the two sound sources in a stereo-phonic installation, one of the loudspeakers beingthe other. For the sake of simplicity we assume thatthe observer is in the plane of symmetry of the twosources (if he is otherwise situated, the values to bementioned presently must be modified by suitableamounts). We then know that he will locate thevirtual sound source (the "sound image") in theorchestra if (a) the sound from the orchestraand that from the loudspeaker arrive at the same

time but the intensity of the orchestra is at least10 dB above that of the loudspeaker, or (b) or-chestra and loudspeaker are equally loud but thesound from the orchestra arrives 2 msec earlier.These are ·the two extremes; for intermediate cases,as far as the position of the sound image is concer-ned, a lag of 1 msec in one sound is compensated byan increased iI~tensity of 5 dB, according to the al-most linear relationship 9) shown in fig. la. Thesound image is still located in tbe orchestra evenwhen the intensity of the orchestra is 5 dB less thanthat from the loudspeaker, provided the sound fromthe orchestra arrives 3 msec earlier.

'5dB

10

5

Fig. 10. Differences in intensity (in dB) plotted against thepbase differences Lit which produce the sarne angular displace-ment of the sound image.

Another condition is that for no listener maythe sound from the loudspeakers arrive with sogreat a time lag that it is no longer experienced asreverberation of the orchestra, but as a separateecho. This means that for no observer may thefirst loudspeaker signal which reaches him arrivemore than 50 msec after the direct sound from tbeorchestra. This value con·esponds to that foundearlier when investigating tbc maximum time inter-val during which, in speech, the indirect soundcontributes to the intelligibility (see Introduction):Recent investigations 10) have confirmed this valueand also that the indirect sound may be strongerthan the direct without disturbing the locationof the sound image.On the basis of these data we can set down the

following rules for a stereo reverberation instal-lation:

0) K. de Boer, Stereophonic sound reproduction, Philipstech. Rev. 5, 107-114, 1940.

10) H. Wallach, E. B. Newman and :'\1. R. Rosenzweig, Theprecedence effect in sound localization, Arner. J. Psych.62, 315-336, 1949; G. Meyer and G. R. Schodder, Db erden Einfluss von Schallrückwürfen auf Richtungslokali-sation und Lautstärke bei Sprache, Nachr. Akad. Wiss.Göttingen, Ha, 31-42, 1952.

l\1ARCH 1956 STEREO REVERBERATION 265

1) No member of the audience may receive soundfrom any loudspeaker before the direct soundhas reached him.

2) Nowhere may the first loudspeaker sound arrivemore than 40 msec after the direct sound(a safety margin of 10 msec has been deductedfrom the limiting value of 50 msec).

3) The intensity of the loudspeakers may nowherebe more than 5 dB above that of the directsound 11) (apart from the diffuseness required,this is another reason why many dispersed loud-speakers must be .used).

If the same hall is also to be used for plays or lectures andimprovement of the intelligibility is desirable, the stereo rever-beration installation can also make itself useful for thispurpose. One must then ensure that no sound is repeatedlater than 40 msec after the direct sound, and no feedbackshould be applied.

Directions for operation of the installation

Complying with the rules listed above does notnecessarily ensure that the installation works satis-factorily. Because the theory lacks sufficient experi-mental backing, one should beware of clinging tounfounded preconceived opinions.An example will illustrate this. In one of our

first experiments with stereo reverberation, it wasthought that the artificial reverberation should bebuilt up of as many repetitions as possible, in orderto obtain the smoothest possible exponentiallydecreasing intensity. It appeared, in fact, thatthough this did give the impression of a long rever-beration time, this was by no means accompaniedby the feeling of being in a large hall - rather thatof a small "hard" room such as a bathroom. Tosuggest a large space, it was necessary to increasethe time interval between the echos and to make thereverberation not at all smooth. By careful adjust-ment in a laboratory room of about 1000 mavolume, we could create the acoustic impressionof being in a cathedral.

One general piece of advice relating to electro-acoustical' intervention in musical performancesis that the engineer must show considerable retraint.He must suppress his desire to make the effectstriking and take care that he does not exaggerate.His is a thankless task: whenever his work is rec-ognized for what it is, he will be repro.ached, and thebetter he does his work, the more natural the resultwill appear and the less thanks he can expect.The highest praise he can receive is probably the

11) See fig. 8 in the article by Meyer and Schodder, referredto in 10).

simple verdict: " ... the orchestra sounded muchbetter".

As we have said, not only the Philips Theatreat Eindhoven, but also the Gebouw voor Kunstenen Wetenschappen in The Hague, is fitted with astereo reverberation installation. On November.'30th, 1954, this installation (then still provisional)had its public debut during a concert given :by theResidentieorkest. After the concert, the effectswhich can be achieved were demonstrated more. .emphatically.

The verdict on the operation during the concertvaried from "favourable" to "very favourable".Some people, however, found it difficult to observethe effect consciously; This is illustrated by aremark from a musician in the audience: "It wasremarkable that one could not consciously hearthat the installation was in operation; one only felt,or experienced it. Only during the demonstrationafter the concert did I become consciously awareof it, and convinced that my observations duringthe concert were not imaginary".. Though the improvement may be appreciatedonly unconsciously by some of the audience, it isanother matter for the performing musicians, bothmembers of the orchestra and soloists; they expe-rience the stereo reverberation very clearly and, co~s-ciously as making the hall more "playable". Thisundoubtedly contributes to the attainment of ahigher artistic level.

We feel that it is an important milestone in thedevelopment of electro-acoustics that leading musi-cians not only permit microphones and loud-speakers and all that goes with them, in the concerthall, but actually welcome their help.

Other applications of stereo reverberation

Stereo reverberation will undoubtedly find otherapplications apart from the conversion of theatres,acoustically speaking, into concert halls. However,the apparatus is so complicated that these appli-cations will be limited to the professional field forthe time being.

One obvious applioation is in broadcastingstudios. Here, stereo reverberation can be a meansof adjusting the acoustics of the studio to the natureof the music or the play to be performed. This can,of course, be done afterwards by adding artificialreverberation to the microphone signal; the echochamber is a device often used for this purpose.But this deprives the musicians of the stimulusof good acoustics and if there is an audience inthe studio, they too miss the full effect. Stereo

266 .' PHILIPS TECHNICAL REVIEW VOL. 17, No. 9

reverberation can overcome these difficulties 12).Another application that springs to mind is in

the cinema. Here stereo reverberation can be ob-tained in the manner described above, with thedifference that the microphone is discarded and therecording head on the delay wheel is now fed withthe signal derived from the sound track of the film.It is even simpler if the cinema is fitted for a systemlike "CinemaScope", i.e. :fitted with -loudspeakersaround the hall and projectors suitable for filmswith more than one sound track; the direct and thedelayed diffuse sound could then be recorded be-forehand on these tracks, so that the cinema hasno need to be equipped with' the delay wheel.

Finally we may mention the "duplication ofconcerts" 13), that is the stereophonic reproduction

12) According to aprivate communication from Dr. J.J. Geluk,Head of the Dutch Broadcasting Union Laboratories,the Union is planning to build two large broadcastingstudios (7500 m3 each) in which the principles of stereoreverberation will be applied.

13) R. Vermeulen, Philips tech. Rev. 10, 169-177, 1948/4.9and R. Vermeulen and W. K. Westmijze, Philips tech.Rev. 11, 281-290, 1949/50.

in one or more "overspill" halls of a concert givenelsewhere. Clearly, diffuse reverberation in the"overspill" halls can considerably increase themusical value of the programme presented.

Summary. A shortage of good concert halls means that musièis often presented in a hall which is less suitable for this .pur-pose, for example, in a theatre. Such halls can be given betteracoustics for music by providing urtificial diffuse reverberation.An installation is described with which such "stereo rever-beration" can be provided electro-ncoustically. It is based on aso-called "delay-wheel", the rim of which is coated with amaterial suitable for magnetic sound recording. A recordinghead records the music performed. A number of play-backheads (say, four or six) around the circumference of the ro-tating wheel pick up the recorded sound with predeterminedtime lags, and separately feed a group ofloudspeakers, mountedthroughout the hall. The intensity of each group is indepen-dently controllable, and the time lags can be regulated by thespeed of the wheel. The sound is picked up by one or two "linemicrophones", each consisting of ten condenser microphoneswith strongly directional characteristics, to reduce the possi-bility of continuous "howl-back".

It has been shown that such an installation can producea great improvement in the musical acoustics of a hall. Someofthe audience experience this only unconsciously, but the per-forming musicians are strongly aware of it as making the hallmore "playable".Finally, some other possible applications of stereo rever-

beration are discussed: in broadcasting studios, in the cinemaand in the "duplication" of concerts.

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

Reprints of these papers not marked with an asterisk * can he obtained free of chargeupon application to the Administration of the Philips Research Laboratory, Eindhoven,Netherlands.

2217: A. Venema: Thermische emissie (T. Ned.Radiogenootschap 19, 283-303, 1954).(Thermionic emission; in Dutch).

After an introduetion to the subject, the authordeals briefly with Schottky's method for calculatingthe work function, the influence of adsorbed layerson the work function, and its measurement on thebasis of the Richardson formula. This is followed bya survey of emitter substances,. with data on thetemperature dependence of the current density. Thethermionic efficiency is considered. The problem ofthe life of oxide cathodes is also dealt with. Theshortcomings of the oxide cathode are mentioned,which have lead to the development of the so-calleddispenser cathodes. Three types of the latter arediscussed: the L-cathode, the impregnated cathodeand the pressed cathode.

2218: D. Kleis: Dynamique de l'enregistrementmagnetique (Onde électrique 34, 753-760,1954) ..

For good sound. reproduction, an adequatedynamic range is essential. A large orchestra has a

dynamic range of about 80 dB; with magneticrecording, a dynamic range of only 70 dB is attain-able, so that it is important to ensure that this isused to the fullest extent. This puts high require-ments on the recording apparatus as well as on thereproduetion instrument. Unless special precautionsare taken, there is the danger with such a largedynamic range that the noise level and distortionare unacceptably high; this is a result of the asymm-'etry of the erasing current but, more important, ofthe A.C. biassing current. With the usual type ofplay-back head, the signal voltage delivered to theamplifier is proportional not only to the amplitudeof the flux in the play-back head but also to thefrequency. This results in the signal voltage havinga dynamic range considerably greater than that ofthe recording itself. To prevent the introduetion ofnoise and distortion in the amplifier, special circuitsmust be used. This article discusses these factorsand the means by which the desired dynamic rangecan be obtained in both recording and reproduction.The methods used include, in particular, the use ofa negative feedback h.f, oscillator for erasing and

l\'IARCH 1956 ABSTRACTS OF RECENT SCIENTIFIC PUBLICATIONS 267

biassing during recording and an amplifier with .frequency-dependent negative feedback duringreproduction.

2219: G. W. Rathenau and G. Baas: Electron-optical observations of transformations ineutectoid steel (Acta Metallurg. 2, 875-883,1954).

More det~iled account of investigations describedin brief in Philips tech. R~v. 16, 337-339,1955/1956.

. .2220: J. M. Stevels: Networks in glasses and other

. polymers (Glass Ind, 35; 657-662, 1954).

The dielectric losses, measured as tan <5, as afunction of temperature are compared for a numberof glasses, fused silica and crystalline quartz. It isshown that the sharp peak in the curve forquartz at low temperatures has its origin in arelaxation mechanism. For fused silica and theglasses examined, tan <5 has a very broad maximumat low temperatures. It is reasonable to supposethat in the latter cases we are concerned with amovement of the oxygen, modified by the networkin which it is' bound. With decrease in Y. (theaverage number of points of contact per tetrahed-ron), tan <5max and also the corresponding decreaseof the dielectric constant e go through a maximumwhereas the activation energy of the relaxationphenomenon goes through a minimum. This can heunderstood as due to the fact that although thenetwork gets looser and looser, the network-modify-ing ions increase in number so that the oxygenbecomes less mobile. Tan <5 for silicones and otherorganic polymers also, shows a .broad maximum atvery low temperatures. Here, however, sharp peaksare also present at about 200 "K. The possibleorigin of these peaks is discussed.

2221: F. A. Kröger and H. J. Vink: Physico-chemical properties of diatomic crystals inrelation to the incorporation of foreign atomswith deviating valency (Physica 20, 950-964.,1954).

If .foreign 'ions of a valency deviating from thatof ions of the base lattice are incorporated, electro-neutrality is maintained. This is effected in a numberof ways, some of which are. already known: Forexample, electro-neutrality may be maintai~ed bythe formation of vacancies, or by the occupation ofinterstitial sites. A second mechanism is that in.which the, incorporation of the foreign ions isaccompanied by a reduction or an oxidation of thebase lattice. It is shown that it is possible to considerthe various ways of maintaining the electro-

neutrality from a general point of view. There area .number of factors which govern 'the way theelectro-neutrality can he maintained, Among theseare the concentration ofthe foreign ions, the tenden-cy of the base lattice to form lattice imperfections,the position of the energy levels associated with thevarious lattice imperfections, and the width of theforbidden zone. Another factor of importance is theatmosphere in equilibrium with the compound. Insuch a way, a third mechanism of maintaining theelectro-neutrality is found. Application of theseconsiderations to CdS as a base lattice gives asatisfactory agreement with experiment.

2222: F. van der Maesen and J. A. Brenkman:.Acceptor activity of copper in Nand P typegermanium of different resistivity (Physica20, 1005-1007, 1954).

Experiments are carried out in which N. and Ptype germanium samples with various resistivitiesor a bar with a resistivity gradient are saturatedwith copper at temperatures of 750 and 810°C. Thenumber of introduced acceptors calculated from the,change of resistivity after quenching, is dependenton the position of the Fermi level. In material thatremains N type after saturation, there is a consider-able increase of the acceptor activity. It is thereforepossible that copper produces extra acceptor levelswith a rather high activation energy.

2223: H. J. G. Meyer: On the theory óf transîtionsofF-centre electrons (Physica 20, 1016-1020,1954).

In the expressions for the prohahility of a radia-tionless transition in F centres, as given by Huangand Rhys and more recently by the present author,certain parameters occur, the numerical values ofwhich have to be found from a comparison of theexperimental and the theoretical absorption spec_-trum due to the corresponding transition. If certainsmall but systematic deviations between theory andexperiment are neglected, values for the radiation-less transition probability are found which are ofa reasonable order of magnitude. It is furthermoreshown in a qualitative way that the deviations may

.. he explained by the fact that the theoretical spec-.trum is determined under the assumption that theCondon approximation is valid. For alkali-halideF-centres this is probably not allowed. .

2224: B. H. Schultz: Surface recombination as afunction of the concentration of chargecarriers in the interior (Physica, 20, 1031-1033, 1954).

268 PHILlPS TECHNICAL REVIEW VOL. 17, No. 9

.Measured values of the surface recombinationrate s of electrons and holes of Pand N germaniumare compared with what is to be expected from theconsiderations of Brattain and Bardeen on surfacestates. There is qualitative agreement: s is indeedlarge for materials of small resistivity, but it depends'on the resistivity to a lesser extent than the theorypredicts. For intrinsic germanium (i.e. germaniumin which neither donors nor acceptors are dominant)which has been etch-polished, the value of s measur-ed was 20 cm/sec.

2225: F. A. Kröger, H. J. Vink and J. Volger:Resistivity, Hall effect and thermo-electricpower of conducting and photo-conductingsingle crystals of CdS, from 20-700 OK(Physica 20,1095-1099,1954).

Short l'eport on investigations which weredescribed at length in Philips Res. Rep. ~o,39-76,1955: see these abstracts No. R260.

2226: G. H. Jonker: Semiconductor properties ofmixed crystals with perovskite structure(Physica 20, 1118-1122, 1954).

Mixedcrystals ofLaMeH03 and si:Mé+03 (Me=Ti, Cr, Mn, Fe or Co)with perovskite structure havebeen prepared in ceramic form. The cornpoundswith the general formula [La1_xSr,J[Me~_:xMe!+]03show a high electric conductivity as the Me-ionsare present in two valency states. Interestingproperties are met in mixed crystals containing' twokinds of metal ions of the transition group. Becauseof the difference in ionisation energies, one mayexpect that only one kind of the metals, in thesemixed crystals exists in two valency states. It ispossible to prepare samples with high restivity,from which it can be concluded that one metal ispresént in the trivalent state and the other one in thetetravalent state, e.g. in [LaO•75SrO•25][Feg:t5Mn~15]03' By the determination of the maxima of thereaistivity in various series of mixed crystals, thefollowing sequence of preference for the tetra-valent state is found: Mn > Cl'> Fe. So~e of themeasurements are complicated by the ferromagnetic

. properties of the materials, which have a stronginfluence on the conductivity. It was thereforedifficult to find the place of Co in the sequence,

R 271: G. Klein: Rejection factor of differenceamplifiers (Philips Res. Rep. 10, 241-259,1955, No. 4).

By analyzing the ordinary triode differenceamplifier it is shown that the rejection factor can bemade arbitrarily large, without the need for pre-selection of valves or stringent mutual equality ofcorresponding circuit elements. Some circuits aregiven which guarantee a high rejection factor, evenwith 10% difference of the corresponding componentsof the two halves. The theory is verified by a numberof measurements.

R 272: J. Volger, J. M. Stevels and C. van Ameron-gen: Dielectric losses ofvarious monocrystalsof quartz at very low temperatures (PhilipsRes. Rep. 10, 260-280, 1955, NO'.4).

Between 14 and 150 "K the tan c5 vs T curve ofq~artz, measured at frequencies of 1 or 32 kc/s,may show a variety; of maxima. The relaxationphenomena involved are correlated with bothprimary and radiation-induced lattice defects.Results of experiments with clear quartz, artificiallyirradiated quartz, natural smoky quartz andamethyst are reported. A discussion related to thenature of a number of lattice imperfections is given.

R 273: A. van Weel: Phase-linear television recei-vers (Philips Res. Rep. 10, 281-298, 1955,No.4).

A television receiver with phase-linear int~rmedi-ate frequency amplifier is described. Normal select-ivity demands are satisfied and conventional circuitsare used. Performance compares very favourabywith receivers of which the I.F. phase errors arecompensated for in the video-frequency part ofeither transmitter Ol' receiver. The picture quality,apart from being optimum for the give~ bandwidth,is almost independent of tuning variations.