CA1190984A

CA1190984A - Arrangement for avoiding undsirable oscillations in a closed-loop system and loud-speaking telephone set employing this arrangement

Info

Publication number: CA1190984A
Application number: CA000402425A
Authority: CA
Inventors: Yves J.F. Hetet; Gilbert M.M. Ferrieu; Jacques Y. Balch
Original assignee: Telecommunications Radioelectriques et Telephoniques SA TRT
Current assignee: Telecommunications Radioelectriques et Telephoniques SA TRT
Priority date: 1981-05-08
Filing date: 1982-05-06
Publication date: 1985-07-23
Also published as: AU8346982A; DE3261682D1; EP0065335A1; AU547891B2; JPH0662104A; EP0065335B1

Abstract

ABSTRACT:

This arrangement has for its object to avoid sustained annoying oscillations in a looped system, such as a looped electro-acoustic system or a control system.
It comprises a variable-gain amplifier (11) incorporated in a first path (C1) which forms part of the loop.
Connected between the input terminal (A) and the output terminal (B) of the said first path is a second path (C2) having at least one part which is separated from the first path. Amplifier (14) is provided to ensure that the gain in the second path remains higher than the gain in the first path in the overall frequency band where the oscillations are liable to be produced. A signal (W) formed in the said part of the second path which is separate from the first path is applied to the input of a linear regulator 13, which controls the variable-gain amplifier (11) of the first path and variable-gain amplifier (12) of the second path so as to keep the sig-nal at the input of the regulator (13) constant from a certain value of the signal (W) formed in the second path.

Description

8~
PHF 81.543 C 1 23~3,1982 "Arrangement for avoiding annoying sustained oscillations in a closed~loop system and loud-speaking telephone set employing this arrangement".

The invention relates to an arrangement for avoiding annoying~ sustained oscillations in a closed-loop system, the arrangement comprising a variable~gain ampllfier incorporated in a first path which forms part of the loop o.f the said system.
Among the closed-loop systems in which anno}~ing sustained oscillations may be produced there are~ for example~ systems for controlling electrical or n1echanical or physical quantities. It is known that during the design of such systems, the gain and phase conditions in the control loop must have proper attention paid to them in order to avoid sustained oscillations in the loop. But there is never a certainty that due to unforeseen circum-stances, for example parasitic couplings ~ sudden pertur-bations, self-oscillations will not be produced which may .
attain high and even dangerous amplitudes in certain systems. A different type of closed loop system is an electro-acoustical system which comprises~ for example, a microphone and a loudspeaker which are coupled to any arbltrary manner by means of an electric circuit; if on the other hand these two transducers are coupled by mea~s of an acoustic circuit 7 an electro-acoustic loop is obtained in which oscillations of a high amplitude may be produced which cause unacceptable how~sO This phenome- I
non9 which is known as the Larsen effect, may be produced in, for example, sound reproduction equipment or in loudspeaking talephone sets.
To avoid oscillations in an electro-acoustical loop the procedures which are generally employed consist in the pro~ision in the loop of at least a ~-ariable amplifier or attenuator circuit, which are controlled in accordance with different criteria in order that the gain ., '~

8~
PH~ 81.~l~3C - 2 23 3.19~2 in the loop re~ains below unity. Thus, a procedure emplo~-ed in loudspeaking telephone sets consists of checking, by means of two envelope detectors, ~ihether the speech signal is present in the loudspe~ker path 5 or in the microphone path and producing a gain increase in the useful path and a gain reduction in the other path.
- These gain modifications which follow each other in both paths during a conversation are very unpleasant for the listeners and in addition this type of procedure is not lO ~-ery effecti~e ~or the important acoustical couplings, t as the gain control does not substantially depend on the coefficient of acoustic coupling.
Another procedure described in ~rench Patent Application No 2~461,412 also employes two signal en~elope tS detectors in the two paths and a single amplifier in the loudspeaker path, the gain of which is varied by a diffe-rence signal between the output signal of the envelope detector of the microphone path and the output signal of the envelope detector of the loudspeaker path9 weighted 20 with a predetermined coefficient. For distances which are lo~ger than a predetermined minimum distance between the loudspeaker and the mic~ophone9 the gain of the amplifier is in a constantDatio to the coefficient of acoustic coupling and the oscillations can only be a~oid~d by means 25 of this procedure below said minimum distance. In additiong the use of envelope detectors causes the recep-tion to depend on the speech signals and particularly on the speech signal produced by a speaker who speaks into the microphone.
The pres2nt in~ention provides a different means to avoid annoying loop oscillations 9 by mitigating the disad~antages and the limitations of the prior art procedures. The in~ention not only relates to electro-acoustic loops, but also to all types of closed-loop s~--remC such as control systems, The in~-ention is based on the idea of causing said loop oscillations 9 when the loop oscillation -PHF 81.543C 3 23.3~1982 conditions have been created, to pass through a second path which has at least a distlnct porti.on of the first path and controlling said osc.illati.ons b~ rneans o:f a regulator so as to bring them to a constant level which is low and not annoying in the remaining part of the loop.
According to the invention, a second path which has at least one part which is separate from the first path is formed between the input and output terminals of the first path, means being provided to ensure that the g~in in the second path remains higher than the gain in the first path, in the overall frequeney band where the oscillations are liable to be produced, a signal formed in the said part of the second path which is separate from the first path being applied to the input of a linear regulator controlling at least one variable-gain ampli-fier which forms part of the second path to ensure that the signal at the input of the said regulator remains constant from a certain value of the said signal formed in the seoond path.
In certain embodiments of the inven-tion the second path is completely separate from the first path and comprises a variable-gain amplifier whose output sig~al is applied to the input of the said linear regulator, the latter simultaneously controlling from the input terminal which is used jointly by the two paths the said variable-gain amplifier ~ the second path and a variable-gain amplifier provided in the first path ar~d/or a variable-gain amplifier provided in the first path from the output terminal which is jointly used by the two pathsO
In further embodiments of the inven~on it is possible to emit one variable-gain amplifier? because of the fact that the first and second paths comprise one common variable-gain amplifier which is provided at the input terminal which is used jointly by the two paths and/or a variable-gain amplifier provided at the output terminal which is used jointly by the two paths, the one or the other of these variable-gain amplifiers jointly used by ~ 4 PHF 81.543C - 4 ~3.3~1982 the t~-o paths being controlled by the said linear re~ulator.
~d~-antageousl~r, the non-annoying oscillations t occurring in the loop which is clos~d by the second path 5 are produced at a well-defined frequency. This result is preferably obtained by means of a ~ilter having a narrow passband which is connected into that part of the t second path which is ~Eferent from the first path, before the spot where the signal is taken off which is applied to the input of the linear regulator, 30 as to selectively increase the gain in the second path.
In addition~ in order to satisf~ the gain con~
ditions relative to the t~o paths it is possible to include a further filter in that part of the first path which is different from the second path~ this filter increasing the gain in the same narrow passband as the filter of the second path. It is alternatively possible to provide such a filter before the input terminal which is used jointly by the two paths, after the output termi-~al which i5 used jointly by the two paths or in the parts which are used jointly by the two paths.
The arrangement in accordance with the in~ention can be advantageously employed in a loud$eaking telephone set, the coupling circuit coupling the set to the telephone line then being included in that part of the first path which is dif~erent from the second pathl I~ tke loop which will be denoted auxiliary loop hereinafter and which comprises the acoustic patn between the loud~eaker and the microphone of the telephone set and which is closed by the second path, sustained amplitude oscillations are obtained which are controlled by the regulator to ensure that they are non-annoying and~ in particular, inaudible, al the same time~ annoying Larsen oscillations are pre~en-red from being produced in the electro~acoustic loop9 denoted main loop, ~hich is closed b~- the first path, these annoying oscillations being produced o~-ing to imper-fections in the coupling circuit coupling the local sub ~ 8 ~
PHF 81.543C 5 23~3.1982 scriber's set or owing to an acoustic ooupling in the remote subscriber's set.
~ The arrangement in accordance with the in~ention when us0d in a loudspeaking telephone set of the "hands free" type renders it furthermore possible to effect in a very simple manner an automatic change~over of the gains in the transmis~ion and receiving path of the telephone set. In this respect it should be noted that in many prior art "hands free" telephone sets~ for example the 10 sets disclosed in French Patent Application No~ 2~376,576, the gain change-over by speech is employed in order to pre~ent annoying Larsen oscillations from being produced.
In teLephone sets of this type the speech si~lals in the transmission and receiving paths are detected to determine i~
15 the path in which the speech signal is the higher; a command to increase the gain is effected in this path and a gain to decrease the gain is effected in the other path, these two gain variations being complementary so as to ensure that the total gain in the loop where the Larsen 20 oscillations may be produced remains below unity.
In a telephone set of the "hands-free" type provided with the arrangement in accordance with the invention it is not ~ecessary to effect a gain change-over b~ the speech to prevent annoying Larsen oscillations ~5 from being produced, but it may be usefulg for other reasons~ to effect a manual gain change-over9 or a change-over controlled by the speech signal. For example~
when the "handsfree" s~ is operated in a very noisy environment it may be important to increas& the gain 30 in the receiving link in order to increase the listening comfort~ and to decrease the gain in the transmission link9 even if this may make it necessary to speak louder or closer to the microphone. This gain change-over may be made operati~e manually or controlled directly by the speech.
35 In this second hypothetical case, when the speaker does not speak before the ''hands free" set, the gain in the recei-~ing link must be increased9 which increases the listening g8~
PHF 81.543C ~ 6 23 3.1982 comfort, and the gain in the transmission link must be decreased, ~-hich also prevents the ambient noise of the localit~- in which the "hands free" set is placed from being heard comparative].y loud. In contrast therewith9 ', s ~-hen a speaker spec~s i~ltO the "hands free" set~ the gain in the transmission and receiving li~cs may have the t - nominal values again.
It has surprisingly been found that the arrange-ment in accordance with the invention renders it possible 10 ro emplo~ an extremely simple means to effect these gain--change-overs in the transmission and r0ceiving paths of a relephone set.
In accordance ~-ith a first alternative~ in a relephone set incorporating the arrangement in accordance 15 ~-ith the invention a~d comprising in the ~irst path a ~riable gain_amplifier contro~ed by the linear regulator and arranged in the receiving path, a further variable-gain amplifier having for its object to changeover the gain in the two paths is pro~ided in the receiving path bet~een 20 the microphone and the inpu-t terminal which is jointly used b~ the two distinct parts of the first and second paths~ the said means provided to ensure that the gain in the seco~d path remains higher than the gain in the first path being adjusted in such a marn~r that the electro-25 acoustic loop which is closed by the second path will perma~entl~ be the source of unannoying oscillations of a lo~ amplitude~ a variation caused by the gain of the said further variable-gain amplifier determining a gain ~riation in the opposite sense to that in the receiving 30 path.
In accordance with another alternative~ if ~he variable-gain amplifier which is included in the first pa.h and is controlled by the linear regulator is provi~
ded in the transmission path, the variable-gain amplifier ch has for its object to change over the gain is p-o~-ided in the receiving path between the loud-speaker and ~he output terminal which is used jointly b~ the PHF ~1~543C 7 23~3.~9~2 distinct parts of the first and second paths.
In the ~ariants of the arrangement in accor-dance ~-ith the in~ention which emplo~- a ~-ariable-gain amplifier which is controlled by the linear regulator 5 and is used ~ointly b~ the first and second paths 9 accurately complementary gain ~riations are obtained in the transmission and recei~-ing paths by controlling a gain variation only in the amplifier which has for its object to change over the gain and which is situated either in 10 the transmission path or in the receiving path. ,;
In the two alternatives for the gain change-o~-er system it is possible to control the gain ~-ariation of ~he gain change-o~er amplifier to effect an automatic gain change-o~-er in the two paths by employing a detector for 15 detecting the speech signal from the microphone.
The following description given by way of e~ample with reference to the accompanying drawings will make it better understood how the invention can be pui into effect.
Figure 1 sho~s the arrangement in accordance with the inYention used in an loudspeaking telephone set.
Figure 2 shows the characteristic of the r~gulator employed in the arrangement in accordance with the invention.
Figure 3 shows the diagrams 3a to 3f corTespon-ding to the variations in the arrangements in accordan~e ~ith the in~ention.
Figure 4 shows the arrangement in accordance with the invention used in a control system.
Fi~ure 5 shows an embodiment of the regulator and the two ~ariable-gain amplifiers 9 controlled b~ the regulator.
Figure 6 shows the arrangement of Figure 1, additionall~- pro~-ided with means in accordance wi~h the 3~in~-enrion for changing o~-er the gains in the two paths o~
a telephone set.
1, 89~
PHF 81.5~3C 8 23.3 1982 Figure 1 sho~-s an electro-acousticalloop, formed, for example, by a loudspeaking telephone set.
~aid telepho.ne set comprises a coupling circuit 1 ~-hich couples the telephone line 2 to the transmission path 3 of the set which i5 provided with the microphone 4 and an amplifier 5, and also the receiving path 6 of - the set which is provided with the receiver amplifier ~ and the loudspeaker 8. Between the loudspeaker 8 and the microphone 4 there is a certain degree of lO acoustic coupling, which in particular depends on the distance between these transducer~ and their orientation and which may vary in telephone sets of the type usually referred to as "amplified reception", the microphone of ~-hich, ~hlch forms a fixed part of the combination, has an 15 indeterminate position with respect to the loudspeaker.
This c~upling may be characterized by a coupling coefficient ~ ~ being the attenuation of the acoustic po~-er transmitted by the loudspeaker and reaching the microphone.
Moreover, due to inevitable imperfections in the circuit 1~ the signal appearing in the transmission path 3 of the set is not wholly retransmitted over the telephone line 2 and a fraction of this signal is found in the receiving path 6 of the setO A further inadvertent ~5 coupling be-tween the transmission and receiving paths of the local set shown in the Figure may also be produ~sd by the path incorporati~g the telephone line 2~ ~hich is connected to a remote loud-speaker in which the~e also exists an acoustic coupling between the loudspeaker and 30 the microphone r Whatever the origin of the inadvertent coupling between the transmitting path 3 and the receiving path 6 of the telephone s~t it is possible to establish a device 9 arranged bet~-een the transmission and receiving paths of the telephone set~ and having a defined gain G1 3v ac ~he standard of rhe rario bet~-een the signal appearing on rhe transmission path 3 and the signal appearing on the receiving path 6 of the setO No change is made to ~ 8 PHF gl.543C 9 23.3.1982 the generality of the definition of said device 9 if a circuit 10t for example an amplifier, is arranged in series with the coupling circuit 1, the gain &,l then including the ga.in of said circuit 10.
The acoustic coupling between the lo~speaker ~ and the microphone 4 and the unwanted coupling pro~
duced in the device 9 between the transmission path 3 and the receiving path 6 of the telephone set forms an electro-acoustic loop in which oscillations may come into being if the gain of the loop is equal to or higher than unity. The fr0quency of these oscillations is not accurately known and ls situated in the passband of the components ~hich are included in the loop, that is to sa~
approximately 300-4000 H~, The ampli*ude of these oscil~
lations is only limited by the taturation of the components of the loop.
In order to avoid these unwan~d loop oscilla tions which render the use of the telephone set substan- ¦
tially impossible, it is known to include a variable-gain amplifier 11 in the receiving path 6 of the set and to control this gain g1 from the envelope signal of the signals of the transmission and receiving pathsO This procedure of controlling the gain is not adequate for a hi,~h acoustic coupling and causes t,he.gain g1 to depend on the useful speech signals.
The invention proceeds in a quite different way9 rendering it possible to avoid these disadvantages According to the invention~ connected to the terminals A and B of a first path C1 of the loop incorporating the device 9 and the variable~gain ampl~ier 11 there is a second path C2 incorporating a further variable-gain amplifier 12, The gain g2 of the amplifier 1~ is controlled by a linear regulator 13 which receives as its input the signal W produced in the second path at the output o~ the amplifier 12 and which produces a gain-control signal Vg2 in such a way that the level of the signal W is kept constant beyond a certain value of the PH~ 81.543C 10 23.3.1982 input signal x of the amplifier 12.
Figure 2 sho~s the characteristic of the output signal 1; as a f~ction of the input signal x of the ampli-fier 12 ~hich is so controlled. Up to a value x of the input signal x~ for which the output signal has a ~-alue W0, the gain g2 of the amplifier is constant - and equal to a maximum value g2M = ~ which ls o deiermined by the slope of the line OC. Beyond xO the level 10 Of the output signal W remains constant and equal to W .
For any arbitrary value x of the input signal the operating point of the amplifier est~blishe~ itself at M
and the gain of the amplifier g2 = decreases 1~hen x increases, The important fact is stressed here that with 15 the linear regulator l3 associated with the amplifier 12, the last-mentioned amplifier does not introduce any non-linearity in the second path.
The amplifier 11 in the first path C1 is controlled by a gain-control signal vg1 supplied by 20 the regulator 13 in such a way that its gain g1 responds to the gain g2 of the amplifier 12 in the second path C2.
~t each instant the gain g2 Qf the amplifier 11 may be equal or proportional to the gain g2 of the amplifier 120 If the two amplifiers 11 and 12 are of the same construc~
25 tion~ the t~o control signals vg1 and vg2 may be equal and ma~ be obtained from the same output of the regulator as sho~ in the Figure. Fin~lly~ means are provided in the one and/or in the other one of the two paths 50 as to ensure that the gain in the second path C2 remains 30 higher than the gain in the first path C1 in the overall fre~uency band where the oscillations are likel~ to occurO
Tn Figure 1, these means are represented by the amplifier 1~ ha~ing a gain G2~ which is arranged in the second path C~ before the variable-gain amplifier 12 and is preferably ~, _ 1 ~ accociated ~-ith an attenuator 15 having a gain -A ~ which is connected bet~-een the output of the amplifier 12 and rhe outtput rerminal B which is used jointly by the two PH~ 810543 ~ 11 23~3.1982 t paths. The abo~-e-described notations being used, this gain condition in the t~;o paths ma~ be ~-ritten aâ: ¦
~ t:
G2 . A2 ~ g2 > Gl gl ( 1 ) It should be noted here that this condition (1) must be realized for every possible frequency of the loop oscillations t that is to say~ in the e~ample chosen here, for ever~r ~requency in the band 300-4000 Hz~ r In the event that the two amplifiers 11 and 12 are identical and controlled by the same signals, condition (1) beco~nes:
G2 ~ 2 > G1 lS To explain how the system of the invention operates and to demonstrate that it is possible to avoid annoying loop oscillations therewith it is advantageous to define the gain G of an electro-acoustic circuit ~-hich is enclosed by the frame 6 and comprises the loudspeaker 8 and its amplifier 7, the microphone 4 and its a~plifier 5, a~d finally the acoustic path between the loudspeaker and the microphone~ defined by the coupling coefficient ~ .
This gain G may be defined as the coefficien~ ¦
of the ratio between the signal v applied to the ampli- ¦
fier 7 and the signal u produced by the amplifier 5.
Several quantities may be defined to e~press said gai~ Go Th~ acoustic power up to the output of the loudspeaker is denoted ~ and the acoustic po~er just be~ind the microphone is denoted q. Obviously it is then apparent that:
P ~ 2 p To introduce the transduction factor of the loudspeaker it is possible to define a nominal acoustic po~;er p cupplied by the loudspeaker 8 and ~;hich ~ould be producec b~- a signal ~- at the input. of the amplifier ,. The transduction factor of the assembly formed b~- the amplifier 98~

PHF 81.543 C ~ 12 2303.1982 7 and the loudspeaker 8 is ~ v ~ In the range iIl ~-hich said amplifier-loudspeaker assembly is linear~ it mav be written that:

[ 0] (3) - To introduce the transduction factor of the microphone it is possible to define an acoustic power qO measured ,.
just before the microphone and produced by a speaker lO ~-hich produces an acoustic power Kp (K being a constant) and situated at a nominal distance characterized by a coupling coefficient ~1 , A signal u at the output of he amplifier 5 corresponds to this acoustic power qO -~O The transduction factor of the assembly formed by the microphone 4 and the amplifier 5 is ~ ~ = [K
In this range in which said microphone~amplifier assemblyis linear, it can be written that:
[ Kpo ~ o ] (4) Taking account of the relations (2)~ (3) and (4) the gain G of the electro-acoustic circuit 16 can be brought to the form~ ¦
p u u o o G = ~ ~ ~ vO ~
Sin~e, in a~cordance with the oondition (1) to be satis~ied7 *he gain in the second path C2 between 3~ A and B is higher than the gain in the first path C1 oscillations can only be present in the auxiliary loop formed by the electro~acoustic circuit 16 and the second parh C2 and cannot establish themselves in the main loop formed by the circuit 16 and the first path C1 . Let it first be ass~med that in said auxiliary loop ~here is no speech signal produced b~r the microphone 4 or coming from the telephone line 2. There are no oscillations if the ~9~9~ -P~ 81.543 c 13 2303. 1982 i.
gain in this auxiliary loop is belo~ it~-~ that is to sa~ if: 1 G . G2 . ~2 ~ g2 <-1 (6) In this inequality, the gain g2 of the amplifier 12 must be given its maximum value g2M~ deflned in the ~oregoing.
As in accordance ~ith formula (5) the gain G is propor- !
tional to the coupling coe~ficient ~ between the loudspeaker and the microphone ? the inequality (6) simpl~r confirms that there are no oscillations if said coupling coefficient is lower than a predetermined ~alue.
If the coupling coef~icient between the loud- f speaker and the microphone increases~ the gain G increase~
and abo~e a predetermined value of the coupling coefficient the inequality (6) is no longer satisfied. This results in oscillations starting in the the auxiliary loop formed b~ the circuit 16 and the second path C2. The linear regulator 13 causes a decrease of the gain g2 of the ampli~ier 12 in such a way that the amplitude of these 20 oscillations at the output of the said ampl fier remains limited to the value WO. The output signal o of the attenuator circuit l5 then has a value which is s~ffi-ciently low to ensure that all the components of the loop, particularly the ampli~iers 5 and 7 of the circuit -¦
16, operate in the line~r mode. For this oscillating mode the total gain of the loop establishes itself at the ~alue t, that is to say it can be written that:
G . G2 . A2 g2 ~ (7 Since the components o~ the circuit 16 operate ln ~he linear mode, the formula (5) which furnishes the gain G may be employed and by combining the formulae (5) and (7) it is easily obtained that:

g2 ~ ~G~ ' uO ~ ~8) This formula (8) sho~rs that ~or this oscillating mode in the auxiliary loop formed by the circuit 16 and ~ 9 ~ ~
PHF 81.543 C ~ 14 23 3 1g~2 the second path C2 the gain G2 of the amplifier 12 establishes itself at a value which is proportional to the acoustic attenuatio~ :~actor ~ bet~v-een the loud- j spea~er and the nlicrophone~ which itself is substalltially proportional to the distance between the two transducers.
The gain of the amplifier ~11 in the first path, ~-hich responds to the gain g2 establishes itself there fore automatically also at a value which is proportional to the acoustic attenuation factor ~ ~ that is to say l substantially proportiona]. to the distance bettveen the loudspeaker and the microphoneO This gain g1 varies in the same sense as the gain g2~ whic}l renders it possLble to satisf~ the conditioll (1) in all circumstances. This condition (1) being satisfied~ the total gain in the main lS loop formed by the circuit 16 closed by the first path Cl always remains below lln;ty~ so ~at this first path Cl does not contribute to the production ofbigh-amplitude and anno~ing Larsen oscillations at the terminals of the electro-acoustic circuit 16.
Thus, with the device in accordance with the in~ention, when the acoustic attenuation factor becomes sufficiently low to create oscillating conditions in the loop these oscillations are forced to pass through the second path C2, particularly because the gain g1 of the 25 amplifier 11 of the first path C1 responds to the gain g2 of the amplifier 12 of the second path. In the seço~d path the loop oscillations are controlled in such a way that they are not annoying in the electro-acoustic circuit 16, that is to say they are inaudible and do not saturate 30 the amplifiers 7 and 5; thus) for a given maximum amplitu-de W of the oscillations at the output of the amplifier 12 it is always possible to increase the attenuation factor A2 of the circuit 15 so as to ensure that the amplitude of the oscillations at the input of the circuit ~ ~ is extremel~- small and not annoying~ the gain G2 of .he circuit 14 being increased in a corresponding ~a~- to sa~sf~ the condition (1) The signal produced at the output PHf 81.543 C 15 23~3~1982 terminal B b~- the oscillations ci~culating in the second path C~ has the fi~ed value ~rL = WO
Let no~ the case be considered ln which the first path C1 and the electro-acoustic circuit 16 are the source of the usef~l speech signals. Let it first t be ass~ed that the speech signal produced by the micro- s phone 4 is totally directed, starring from point A~ along the first path A1. Loop oscillations may then come into e~istence in the-second path C2, e~actl~ the same as des-cribed in the foregoing9 without being affected by the speech signal coming from the microphone. At point B~ ,~
these loop oscillations have a very small amplitude and superpose themselves on the speech signal coming from the telephone line 2 v~a the output of the first path C1.
A listener in front of the loudspeaker 8 will not observe these loop oscillations of a very small amplitude.
In the h~pothetical case posed above t the gain g2 f the amplifier 12 and consequently gl of the amplifier 11 20 of the first path are not influenced by the speech signal coming ~rom the microphone 4 and only depend on the acousti~ coupling coefficient ~ .
Actually~ if no p~ecautions are ta~en9 th~re is a risk that a fraction of the signal coming from the t 25 microphone 4 is applied to the second path C2 and disturbs the regulation of the amplitude of the loop osci~ations by means of the regulator 13. In that event~
the gain g2 of the amplifier 12 and consequently the gain g1 of the amplifier 11 will not only depend on the ', 30 acoustic coupling coefficient, but also on the amplitude of the speech signal coming from microphone 4O A means to avoid this disadvantage is to provide a filter 17 which is indicated by means of dotted lines in the Figure in the second path C2 before the amplifier 12. The gain 35 G2 in the second path then includes the gain of the ampli-fier 14 and the filter 17.
If, in the first path C1, the gain for the frequencies above, for example, 3OOO Hz is higher than ~ ~ ~9~ ~ 8 ~

PHP 81.5~3C 16 23.3,1982 the gain for the frequencies~ower than 3OOO Hz, it is possible to use a high-pass filter as the filter 17, ~;hich high-pass filter lncreases the gain in the second path for the frequencies higher than 3OOO Mz. The condition (1) is thus easily satisfied and the speech signals affect the regulation procedure of the loop - oscillations to a lesser extent. These loop oscillations occur at a frequency above 3OOO Hz but still poorly defined, ~hich is a disadvantage.
It would be more advantageous if the filter 17 ~ere a bandpass filter having a very narro~ passband~
~is filter selectively increases the gain G2 in its pass-band and establishes an adequate phase shi~t so as to ensure that the gain in the auxiliary loop formed by the circuit 16 and the second path C2 may take a positive value in the said passband and that the oscillations in this au~iliary loop establish themselves substantially at the c~ntral~requency of the said passband~ independent of the acoustic coupling between the loudspeaker and the 20 microphone; at the same time the speech signals in the second path have a ver~r low level compared with the lo~p oscillations and do ~ot substantially disturb the regula tion of the amplitude of~hese oscillations.
It should here be noted that ~ith a filter 17 25 having a ~arro~ passband it becomes much easier to satisfy the ~ndition (1) which ensures that the loop oscillation~ pass only ~ia the second path C2~ To satisfy this condition it is sufficient that the gain of the seoond path fGr the central frequency of the narrow band 30 of the filtsr 17 is superior to the gain provided by the first path for all the frequencies of the band of said first path (for example 3OO-4OOO Hz). It becomes still easier to satisfy this condition (1) by providing a filter 18, ~-hich is represented by the dotted line in the first ~ pa~h, increasing the gain in the same narro~- passband as ~he filter 17. The gain G1 in the first path then includes PHF 81 o543C 17 23~30 1982 the gain of the device 9 and the filter 18. This filter 18 selectively increases the gain G1 in its passband, so that the gain of the first path C1 provides a maximum value in said narrow passband of the filter t 5 18, To satisfy the condition (1) and consequently to ensure that the loop oscillations pass only along the second path C2 it is then sufficient that the gain of the second path C2 be higher than the gain of the *irst path C1 for the common central ~requencies of the pass-lO bands of the filters 17 and 18 ~ It will be seen here that the filter 18 causes substantially no perturbations in the speech signals~ because of its narrow band. It will be ob~rious that the amplifier 14 and the filter 17 on the one hand and the arnplifier 10 and the filter 18 on the 15 other hand may be in the form o~ two selective ampli~iers, ~inally~ instead of employing a ~ilter 18 in the first path C1 it would alternatively be possible to employ a filter 19, which is represented by dotted lines and is provided in the path between the output of the amplifier 20 5 and the input terminal A which is used jointly by the two paths. ~s said filter 19 increases the gain in the same narrow passband as the filter 17~ the condition (1) is of necessity satisfied automatically arld the loop oscillations at the central frequencies of the two 25 filters 17 and 18 pass via the second path~ ~inally~ it will be obvious that in the above-described variant in which the filter 17 o~ the second path is ~a high-pass filter which increases the gain G2 for the frequencies higher than for example 3OOO Hz, it is alternatively 30 possible to employ a filter 18 or a high-pass filter 19, but with a small attenuation difference between the transmitted high-frequency band and the attenuated lo~-~
~requency band, in order to prevent the speech signals from being distorted to a consi~rable e~tent.
Several different variations of the arrangemen~
in accordance with the invention are possible which will now be described with reference to Figure 1 which shows PHF 81~543C 18 23.301982 circuit diagrams corresponding to some of the possible ~-ariations. In these circuit diagrams the components sho~ .in ~he circult diagram of ~igure l ~hich are essential and sufficient for an ~mde.rstanding of the description are given the same reference numer~s, namel~
the electro-acoustic circuit 16 having a gain G, the de~-ice 9 having a gain G1 proYided in the first path C1, the device 14 having a gain G2 and the attenuator circ~t ha~-ing a gain ~ provided in the second path C2 and ~
finall~- the linear regulator 13. Also the ampli~iers 11 and 12 ha~ing the respective gain g1 and g2 ma~r be used, as ~ill be described herei~lafter. For the sake of` simplici-r~- ~he f`ilters 17~ ~8~ 19 which are shown bv mealls of dotted lines in Figure 1 have been omitted.
In the arrangement shown in ~igure 3a, the t~-o paths C1 and C2 connected between the terminals A and B are absolutely separatey as in the arrangement shown in the Figure 1. The difference between the arrangement sho~n in the Figure 3a and the arrangement of Figure 1 is that the variable-gain amplifier of the first path is provid0d bet~ee~ the input termina7 A which is ueed jcintly by the t~o paths and the de~ice 9 9 instead of being provided bet~e~n the device 9 and the common output terminal B. In its ne~ positiony this variable~gain amplifier of the first path is glven the re~rence numeral 11' and has a gain g910 It is controlled in exactly the same manner as the amplifier 11 of the ~igure 1. As regards the control of the Larsen oscillations, everything described for the arrangement of ~igure 1 is also wholly valid for the arrangement show~ in ~igure 3a, In particular, the Larsen oscillations are limited to a small and unan~o~ing amplitude ha~ing at the terminal B the fixed value ~-L = A 9 ~ being, as explained ~-ith reference to ~~ i--~re 2, rhe maximum fixed ~alue of the signal ~
applied b~- Ihe ~-ariable-gain amplifier 12 to the input of Ihe linear regulator 130 PHF 8~ 543C 19 23.3.1982 In a further variation, not shown, the arrange~
ments shown in Figures 1 and 3a can be combined by emplo~-ing t~o variable-gain amplifiers 11 and 1ll provided in the first path on both sides of the device 9 and both con-trolled by the signal processed by the regulator 13~
In the arrangement shown in Figure 3b the - two paths C1 and C2 connected between the terminals A and B are not absolute separate. The circuit diagram of Figure 3b differs from the circuit arrangement shown in the Figure 10 3a in that the input of the device 14 o~ the second path is not connected to terminal A but to the output of the uariable-gain amplifier 11 ' . When the terminal ~ is always considered to be the input terminal which is used Jointly by the two paths C1 and C2 the ampllfier 1l~ is used 15 ~ointly by these two paths~ while after the output of the amplifier 11 I to the terminal B the two paihs have separate parts. If measures are taken to ensure that in the separate parts the second path has a gain which is higher than the gain of the ~irst path it is obtained in 20 this arrangement 3b that the Larsen oscillations are only produced in the part of the second path which is separate from the first path and that the amplitude ~f these oscillations is small and unannoying, being limited at the terminal B to the fixed value vL = o . In a variation, not 25 shown~ of the arrangement of Figure 2 3b,the linear regulator 13 may control, in addition to the variable-gain amplifiers 12 and 119~ a variable-gain amplifier7 not shown~ which is provided between the output of the deu~e 9 and the terminal Bo The arrangement of ~igure 3c may be considered to be a further improvement of the arrangement of Figure 3b as it is possible to do without one variable gain amplifier. It is different in that the amplifier 12 of the second path is no longer controlled b~ the regulator 137 35 and consequentl~- has a fixed gain g2 In the arrangement of Figure 3c 9 taking account of the fact that the amplifier 1~' having the gain g'1 ~orms part of the first as well as of the second paths between the terminals A.and B~ it can g84~

PH~ 81.543C - 20 23.3.1982 be said that the regulator 13 acts on the gain g'l of rhe amplifier 11 ' for keeping the signal W ~ormed in the second path at the output of the amplifier li~c 11l, 14, 12 co~stant and equal to WO~ which signal is to be applied 5 to the input of the regulator 13. The amplitude of the oscillations on the terminal B is still limited to a lo~-- and fi~ed ~-alue o . In a variation9 not showny of the ~ 2 arrangement of ~igure 3c the regulator 13 can inter alia 10 con~rol a ~-ariable-gain amplifier provided between the output of the de~-ice g and the terminal B, In the arrangement sho~l in ~igure 3d the t~o paths C1 and C2 connected between the two terminals ~ and B are aiso not absolute separate, 1~hen the arrangen~ent oI`
15 Figure 3d is compared with the arrange~nent of Figure 1 i-t ~ill be seen that they differ in that the attenuator circuit ha~ing a gain ~ is connected between the output of the device 14 and t~e input of the variable gain amplifier 11.
When the terminal B is still considered to be the output 20 terminal which is used jointly by the two paths C1 and C2, these two paths ha~e the amplifier 11 in commonO
The separate parts of the two p~hs comp~ising the arrange-ment 9 for the first path C1 and the cascade arrangement of the de~ice 14 and the attenuator circuit 15 for the second 25path C2 are connected between the terminal A and the i~put of the amplifier 11. It should be noted that in this arr~ugement of Pigure 3d the amplifier 12 is ~ot incorpora-t~d in the second path, A signal is ~ormed in that part of the second path which is separate from the first path~ in 30a place situated between the device ~4 and the attenua-tor circuit 15 and this signal is applied to the input of the regulator 13 by means of the variable-gain amplifier 12~
This regulator 13 controls the gain g2 of the amplifier 12 and the gain g1 of the amplifier 1 in such a manner that the 3--ig-al ~~ coming from amplifier 12 and applied to the input o-^ .he regula~or 13 is kept at a constant ~-alue 1; . It can be easil~- demonstrated that on the terminal B the PHF ~.543C 21 23.3.1982 Larsen oscillations have an amplitude which is limitecl to the value vL = o gl which is a fixed value because ' ~ g2 the gains gl and g2 of the ~mplifiers 1l and 12 are controlled by the same signal. Particul~rl~ when the amplifiers are identical, g1 ~ g2 and the limit value o~ !
-the amplitude of the Larsen oscillations is vL = o In a variation, not shown~ of the arrangement of Figure 3d the regulator 13-may control inter alia a variable-gain ampli~ier which must then be provided between the terminal A and the input of the device 9.
The arrangement shown in Figure 3e i9 a further improvement of the arrangement shown in Figure 3d, an improvement which is comparable to the improvement effec-ted in the arrangement of Figure 3c with respect to the arrangement shown in Figure 3b. The arrangement 3e differs from the arrangement 3d in that the gain g2 of the amplifier 12 is no longer controlled by the regulator 20 13 and consequently remains fixed. The r~gulator 13 acts on the gain g1 of the amplifier 11 which is assumed to be part of the second path, so that the signal ~ormed in the second path bet~een the device 14 and the attenuator ci~cuit 15 and amplified thereafter in the arnplifier 12 having a 25 fixed gain g2 9 is applied to the input of the regulator with a fixed amplitude WO. In the arrangement of Figure 3e the amplitudes of the Larsen oscillations on the terminal B ha~e the value vL = o g1 ~ But 9 in contrast with the ~ariations described in2 g2 the foregoing, this value vL
30 is vuria~le and-the gain g2 is fixed. It is nevertheless always possible to choose an attenuation coefficient A2 and a gain g2 in such a manner that the variable amplitude of the Larsen oscillations remains low and non-annoying~
As the amplifier 11 having the variable gain g1 i.s located 35 in the path of the useful speech signals which come frorn the telephone line and are to be applied to the loudspeaker incorporated in the electro~acoustic device 16,it can be said that the variation shown in Figure 3e makes it possible to obtain on the terminal B a constant ratio PHF ~1.543C 22 23.3.1982 ber~een the useful speech signal and the noise generated b~- the unannc)~ring L,arsen oscil:Lations produced b~- means of the second path, In a ~ariation, not S}10~, of the arr~ngement of Figure 3e the regulator 13 might inter alia control a variable~gain a~plifier provided between the terminal A and the device 9~
The arrangement of Figure 3f dlffers from the arrangcment of Figure 3e in that an amplifier 11 t having a variable gain g~1, controlled b~ the regulator l3t is provided bet~een the terminal A and the arrangement 9 and furthermore in that the input of` the device ~4 is co~nec- j ted between the amplifier 1l' and the device 9. Thus~ in this arrangement of Figure 3f the variable-gain ampli~ers 11' and 'I 1~ which are connected to the terminal A and the terminal B7 respectively, are used jointly by the two paths C1 and C2 and arc controlled by the same signal supplied by the regulator 13O The separate parts of the two paths comprise the same elements as the ~rrangement shown in Figure 3e and the regulator 13 has its input connected in the same manner to the secorLd path3 by means of the amplifier 12 having a fixed gain g20 As in the arrangement shown in Figure 3f~ the amplitude of the Larsen oscillations on the terminal B has the value ~ g ~ A g1 which varies in the same way as g1 It can be seen that in the arrangements of Figures 3c~ 3e and 3f where an amplifier 12 having a fi~ed gain g2 is described9 it is possible to omit this amp1ifier completely, if a gain g2 equal to unity is sufficient for a correct opera-tion of the arrangements, In the arrangements shown in Figures 3a to 3f it is advantageous to provide filters having the same characteristics and the same functions as the filters 1,, 18 and 19 in the arrangement sho~n in Figure 1 D A
fil~er ha~-ing the function of the filter 17 would then be p-o~-ided in that part of the second path ~-hich is separate from the first pathl A filter having the function of the filter 18 ~ould then be provided in that part of the .g~

P~ 81.543C 23 23,30 982 first path whlch is separate from the second path, a filter ha~ing the function of the filter 19 would then be di~posed in the par-ts which are common to the main and -the auxiliarr loops~ ~
In the foregoing the use of the invention for t an electro~acoustic system which is inadvertently looped ~!
by acoustic and/or electric coupling is described. How-ever~ the invention is suitable for use in a more general way in all systems in which a loop may be formed in which lO uncontrolled oscillations of a high amplitude may be produced~ So the invention may be used in all types of control systems~ as shown in Figure 4.
Figure 4 shows the well known circuit diagram of a control system intended to have any output quantity 5 S respond to an electric input signal E. To make this more readily understood the same re~erences and the same ~-notations are used as in Figure 1 to indicate the corres-ponding components and quantities. The input signal E is applied ~ia the point A to the forward path of the contrDl 20 system9 comprising the cascade arrangement of the device 9 ha~ing a gain Gl~ the variable-gain amplifier ll and finally the device 20 which supplies the output quantity S. The feedback path of the control system is connected between the output of the de~ice 20 and the point Ao T~is 25 feedback path is represented by the device 16 which has a transfer function modulus G. In response to the output quantity S the device 16 produces an electric signal u which is added with a suitable phase to the input signal Eo During adjustment of this control system or in abnormal 30 circumstances during its operation it may happen that un~anted oscillations arise in the loop formed by the forward path and the feedback path and that the output quantity S attains uncontrolled and dangerously high amplitudes.
The invention renders it possible to a~oid loop oscillations of this typeO By employing, for example, the embodiment shoun in ~igure '~ ~he invention lS utilized bv PHF. 81.543C 24 connecting a second path C2 which comprises in accordance with one of the variations of Figure 1 a filter 17 having a narrow passband and an amplifier 14 which together have a gain G2, an amplifier 12 having a variable gain G2 and an attenua-tor 15 having a gain 12 to the terminals A and B of a first path Cl of the control. loop which comprises the components 9 and 11. The linear regulator 13 re~u-lates the output signal of the amplifier 12 and renders it possible to have the gain gl of the amp.li~ier 11 respond -to the gain g2 of the amplifier 12. Everything which has been described for the electro-acoustic system shown in Figure 1 remains valid. Particularly, the loop oscillations which may come into being are passed along the second path C2 and in said second path their amplitude is controlled by the regulator 13. Thanks to the attenuator circuit 15 the amplitude of these oscillations can be brought to a very low, unannoying yalue at the output S of the control system.
An embodiment of the assembly formed by the variable-gain amplifier 12:and its regulator 13, and by the variable-gain~amplifier 11 is shown in Figure 5.
In accordance with Figure 5, the ampliier 12 which is provided in the second path C2 comprises a npn transistor 22 the emitter of which is connected to a negative supply terminal which serves as a reference, the collector of which receives the input signal x of the amplifier via the series arrangement of the resistor 23 and the capacitor 24 and the base of which is connected to the output of an integrating circuit 25. This integrating circuit 25 receives the pulse-shaped signal Pc formed in the regulatox 13, as w.ill be described hereinafter. The transistor 22 is the adjustable component of the amplifier 12. Actually, on the terminal 26 between the resistor 23 . - ' ,~, g~

PHF 81o543c 25 23~3~1982 t and the capacitor 24 a voltage is obtained which is a variable ~raction of the input signal x of the amplifier, as its level depends on the more or less conductive state ,' of the emitter-collector path of the transistor 22 and 5 consequently on the output voltage of the integrating circuit 25~ It can be easily seen that on the terminal 26 an increase of the output voltage of the integrating circuit 25 corresponds to a decrease of the level of the signal which is available on terminal 26 and conversely, 10 This variable-level signal available on terminal 25 is, for example~ a current im which is applied to a fixed gain amplifier circuit 27~ At its output 28 the circuit 27 produces a var~le current Im ~hich i5 identical to the output signal W of the variable~gain ampli~ier 12. At 15 its output 29 the circuit 27 produces a current Io ~ Im9 wherein Io is a direct voltage having a constant amplitude~
In the regulator 13 the current Io ~ Im is applied to a pulse-width modulator 30, The latter further receives clock pulses from the clock generator 31 and produces a 2n signal Pm ~ormed by pulses the width of which is modulated by the current Io + Im~ The techniques of forming pulse-width modulated pulses are well known, It is useful to state here that the clock pulses are9 in particular~ used to sample the input current of the modulator~
5 When9 at the sampling instants 9 the variable portion Im f this current is zero 9 the pulses of the signal P have a width PO~ Depending on whether the variable curre~t Im is positive or negative at the sampling instants, the pulses o~ the signal Pm have a width which is larger or smaller than PO. Pulses having a width P ~ P 1 and PO - Pm~ 9 respectively correspond to a predetermined level of the variable current I for ~hich this current may take the value I 1 and -I ~. The modulated signal Pm is applied to a pulse-length detector 32, which produces a pulse-shaped signal P formed b~- pulses which are commonly referred to as compression pulses 9 which have a fixed PH~ 8 ~ . ~43C 26 23 . 3 . ~ 982 r duration and are produced each time the pulses of the signal P reach the values P + P 1, The pulse shaped r cignal P is applied to the integrating circuit 25 which t produces a vo~tage whi.ch is representative of the pulse 5 repetition rate of the compression pulses. ~asically, said integrating circuit comprises a capacitor which is charged - b~- a constant current during the duration of the compres-sion pulses and discharged by a current which is lower thar the charging current~ In this ma~ler~ when compression 10 pulses appear ~hich indicate that the level o.f the current Im is e.xceeded, the voltage produced by the integrating circuit 25 increa.ses, which makes the transistor 22 more conductive and produces an attenuation o.f the current le~-el Im, Finall~ the voltage produced by the integrating li circuit 25 stabilizes around an average value which determines a constant level of the current Im. The amplifier 12, ~-hich is associated with the r0gulator 13~ in accor-dance with the embodiment described in the foregoing, has precisely the required characte~istic shown in Figure 2~
20 while the constant le~el ~O indicated in said Figure becomes identical to the constant level at which the current Im is established~
The variable-gain amplifier 11 incorporated in the first path C1 is of the same construction as ampli 25 fier 12, it comprises the same compone~ts designated by the same references9 but provided with ~n accent notation.
Said amplifier 11 is controlled by the pulse-shaped signal Pc which is formed in the regulat~r 13 as described in the foregoingO The transistor 22l9 which is the variable compo 30 nent of the amplifier 11, is controlled by the sa~le signal as the transistor 22 which is the variable component of the amplifier 12. If the components of the two amplifiers, par-r_cularly the transistors 22 and 22', are arranged pair-~ise in a suitable manner, the ~ain of the amplifier 1' is ~~ a~.omatically adjusted at each instant to the same value as .he gain of the amplifier 12.
Since the amplifier 11 has for its object to 9~38~
PX~ 81.543C 27 23.3 1982 process the useful signal (for e~ample the speech signal), and thc ampli-fier 12 to process the oscillation signal of the lltp~ it mayb useful to give the integrating circuits 25 and 25' different time constants~ adapted to $
5 the signals processed by said amplifier On the other hand, by having the pair-wise arranged transistors 22 and ~2' controlled by the same signal, it may be advantageous $o give the fi~ed amplifiers 27 and 27' different gains, the gains of the amplifiers 11 and l2 remaining in a constani lO rario. ~inall~, it will be clear that one sing~e integra-ting circuit may perform the fnnction of the two integra-ting circuits 25 and 25~ having the same time constant.
The circuit diagram sho~n in ~igure 5 is a suitable CiI`CUit diagram to realize the assembly of the 15 amplifiers ~1 and 12 ~nd the regulator 13 in the arrange-ment of Figure 1. But it will be obvious that with the same embodiment, the regulator 13 can control two variable-gain amplifiers which are connected in a different ~anner, or to co~trol only one of these two amplifiers to reali~e 20 the several variations shown in Figure 3.
The present invention9 when applied in a telephone set of the ~'hands free't type to avoid unanno~ing Larsen oscillations makes it also possible to realize 9 in a very simple way~ a further very useful function~ ¦
25 n~mely to realize complementary gai~ variations in the transmission and receivir~ paths of the set~ ¦
In the known tel~phone sets 9 in order to obiain complementary gain variations in its two paths 7 it is necessary to effect ~ith the aid of an amplifier of the 30 transmission path a galn variation in one direction and to effect at the same time with the.aid of a~ amp~fier of the receiving path a gain variation of the same amplitude~ I
but in the other direction. In order to effect a change-over of these gains b~- speech, two speech detectors whose ou~put signals are used to change-over ihe gain are of.en used.
The present invention makes it possible to PHF 810543C ~8 23.3.1982 to obtain e~tre~el~ complementary gain variations in the t~o pa~hs o~ the set by co.ntrolling only the gain of one single amplifier. To describe this ~eature offered by the lnvention~ let~ for e~ample~ the case be S considered of a telephone set which is associated with the arrangements according to the invention as shown in Figure 1 and it will be described how the change-over gains can be effected w:ith re~erence to Figure 6 which shows, referenced in the same wa~9 the majority of the 1n components sho~ in Figure 1. In this embodiment, the single amplifier havin~ for its function to control the gain in the two paths of the telephone set is the ampllf`ier ~; which must no~ be assumed to have a variable gaill and is included in the transmission path between the microphone 15 4 and the input A wh~ch is used jointly by the first path C1 and the second path ~2. If, for the minimum gain value of the amplifier 5, that is to say the minimum value of the gain G the gain o.f the components of the second path C2 is controlled, as described in the foregoing~ in such 20 a ~ay that the sustained~ unannoying oscillations are only produced in the auxiliary loop formed by the circuit 16 and the second path, it is certain that these anannoy-ing oscillations are also produced for all the other values of the gai~ ~ of the amplifier 5~ For all the 25 values of the gain g, the total gain in the auxiliar~ loop remains equal to unity which is expressed by the above fo~mula (7)0 As the oscillations passing through the circuit 16 have a very small amplitude9 the components of this 30 circuit operate in the linear mode and the gain G of the circuit 16 is substantially proportional to the gain Assuming that G = hg, h being a fixed coefficient, the formula (3) may no~- be written:
3 h.g~g2~ 2 A2 = 1 ~or ~he event in ~-hich the two amplifiers 11 and '2, having variable gains g1 and g2 are identical and are controlled by the same signal, it is obtained that PH~ 81~543C ~9 23.3019~2 g1 = g2 and the formula (9) may-be ~Titten~
g-gl'~T2 ~2 = 1 (10) From this it can be deduced that to ever-induced variation of the gain ~ of the amplifier 5 ~hich amplifies the signal from microphone 4 there corresponds -automaticaI~ a variation of thesame amplitude and the opposite sense of the gain g1 of the amplifier 11 which amplifies the signal received in the telephone set~
In the event that the gains g1 and g2 are controlled such that g1 = a g2~ it can be easily seen that to ever~ ~ariation of the gain ~ of the amplifier 5 there corresponds in the recei~ing path a gain ~ariation in the opposite direction and having an amplitude divicled 15 by a.
It should be noted that so as to obtain opposi tely directed gain variations in the transmission and receiving paths an amplifier~ for example amplifier 5, included in the transmission path between the microphone 20 4 and the input terminal A which is used jointly by the two paths Ct and C2 must be acted upon~ An induced gai~
variation i~ any place in the receiving path, for example a gain va~iation of the ampli~ier 7~ will not cause auy gain variation in the transmission pa*hO
The gain variation of the amplifier 5 may be effected progressively or suddenly between two gain ~aluesO
I-t can be controlled for example by a speech signal detector 40~ so as to automatically realize "the change o~er of the speech " in the set. When a subscriber speaks 30 into the microphone 4g his speech detected by the detector 40 serves to con-trol a normal gain of the amplifier 5. This automatically ~esults in a gain g1 ~
a low value for the amplifier 11 and consequently a ~-ea~
energizing signal for the loudspeaker 80 When the 3s subscriber stops speaking into the microphone 4~ the detector 40 controls a lo~ ~ain of the amplifier 5~ This automatically results in the amplifier 5 having a gain g1 PHF 81.543C ~ 3O 23,3.1982 of a higher ~-alue~ which produces a normal energizing cignal for ths loudspeaker 8. It is of co~se possible ~o pro~-ide manual controls for the subscriber, to ensu~e the,best possible listening comfort.
It will be readily understood that the cha~ge-oYer of the ,gains in the two paths may also be effected - in Ihe other variations of the arrangement in accordance ~-ith the invention, by means of one single ampli~ier. If a variable-gain amplifier of the first path C1~controlled lO b~- the regulator 13~ is provided in one of the two paths ``
of the.telephone set it is sufficient for the variable~ ~
gain amplifier used to change-over the gain to be pro~ided ,`
in the other path~ between the acoustic transducer and the terminal which is used jointly by the two separate parts 15 of the two paths C1 and C2. For example, in the arrangement sho~ in Figure 3a~ i~ which the variable~gain ~mplifier 11' controlled by the regulator 13, is provided in the transmission path of the telephone set, the variable-gain amplifier used for changing over the gai~ might be 20 provided in the receiving path between theoutput terminal B7 which is used jointly by the two paths and the loud-speaker incorporated in the electro-acoustic circuit 16.
~or the automatic change--over of the gain by the speech signal it would be possible to emplo~ the speech signal 25 detector 4O of the microphone~ Finally9 it should be noted that i~ the arr~gements such as those shown in the Figures 3c, 3e and 3f in wh~ch the regulator 13 controls only one or two variable~gain amplifiers used jointly b~
the two parts C1 and C2p gain variations of the opposite 30 direction and having automatically the same amplitude are obtained by means of the amplifier changing over the gain pro~-ided in the suitable path.

Claims

THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE PRO-PERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:

1. An arrangement for avoiding sustained, annoying oscillations in a closed loop signal system, the arrange-ment comprising a variable gain amplifier incorporated in a first path which forms part of the loop of the said system, characterized in that between the input and output terminals of the first path there is formed a second path which has, at least, one part which is separate from the first path, means being provided to ensure that the gain in the second path remains higher than the gain in the first path, in the overall frequency band where the oscillations are liable to be produced, a signal in the said part of the second path which is separate from the first path being applied to the input of a linear regulator controlling at least one variable-gain amplifier which forms part of the second path to ensure that the signal at the input of the said regulator remains constant with respect to a predeter-mined value of the said signal in the second path.

2. An arrangement as claimed in Claim 1, charac-terized in that the second path is completely separate from the first path and comprises a variable-gain ampli-fier whose output signal is applied to the input of the said linear regulator, this linear regulator controlling simultaneously the said variable-gain amplifier of the second path and a variable-gain amplifier provided in the first path near the input terminal which is used jointly by the two paths and/or a variable-gain amplifier provided in the first path near the output terminal and common to the two paths.

3. An arrangement as claimed in Claim 2, charac-terized in that the second path is connected by means of an attenuator circuit to an output terminal common to the two paths.

4. An arrangement as claimed in Claim 1, charac-terized in that the first and second paths comprise a com-mon variable-gain amplifier and controlled by the said linear regulator.

5. An arrangement as claimed in Claim 4, wherein the first path comprises inter alia a further variable-gain amplifier which is also controlled by the said linear regulator.

6. An arrangement as claimed in Claim 4 or 5, wherein the signal in that part of the second path which is separate from the first path is applied to the input of the linear regulator by means of a fixed-gain amplifier,

7. An arrangement as claimed in Claim 4 or 5, wherein the signal in that part of the second path which is separate from the first path is applied to the input of the linear regulator by means of a still further variable-gain amplifier, which is also controlled by the linear regulator.

8. An arrangement as claimed in Claim 4 or 5, in which the first and second paths comprise one common single variable-gain amplifier which is provided near the input terminal which is used jointly by the two paths, charac-terized in that the second path is connected by means of an attenuator circuit to the output terminal which is used jointly by the two paths.

9. An arrangement as claimed in Claim 4 or 5, in which the first and second paths comprise a common, single variable-gain amplifier which is situated near the output terminal which is used jointly by the two paths, charac-terized in that that part of the second path which is separate from the first path is connected to the input of the said common amplifier by means of an attenuator circuit.

10. An arrangement as claimed in Claim 4, in which the first and second paths comprise two jointly used variable-gain amplifiers which are provided near the input and output terminals which are used jointly by the two paths, respectively, characterized in that that part of the second path which is separate from the first path is connected by means of an attenuator circuit to the input of the common amplifier which is provided near the output terminal which is used jointly by the two paths.

11. An arrangement as claimed in Claim 1, charac-terized in that a filter is arranged in that part of the second path which is separate from the first path, before the place where the signal is formed which is applied to the input of the linear regulator, so as to selectively increase the gain in the second path.

12. An arrangement as claimed in Claim 11, charac-terized in that the filter of the second path is a filter having a narrow passband.

13. An arrangement as claimed in Claim 11, charac-terized in that the filter of the second path is a high-pass filter.

14. An arrangement as claimed in Claim 11, 12 or 13, characterized in that a filter is arranged in that part of the first path which is separate from the second path, for selectively increasing the gain of the first path in the same frequency band as the passband of the filter of the second path.

15. An arrangement as claimed in Claim 11, 12 or 13, characterized in that a filter is provided and used jointly by the two paths, for increasing the gain of the assembly formed by the two paths in the same frequency band as the passband of the filter of the second path.

16. An arrangement as claimed in Claim 1, 4 or 5, characterized in that the linear regulator comprises a pulse-width modulator which receives a signal which cor-responds to the signal applied to the input of the said regulator, a pulse-length detection circuit which produces a compression pulse each time the width of the pulses pro-duced by the modulator reaches a value which indicates that the level of the signal at its input is exceeded, the said compression pulses being applied to an integrating circuit the output signal of which controls amplifiers con trolled by the regulator.

17. A loudspeaking telephone set using the arrange-ment as claimed in Claim 1, characterized in that a coupling circuit coupling the set to the telephone line is included in that part of the first path which is separate from the second path.

18. A telephone set as claimed in Claim 17, compris-ing in the first path a variable-gain amplifier controlled by the said linear regulator and arranged in one of the receiving path and the transmission parts of the loop, characterized in that a further variable-gain amplifier having for its function to change-over the gain in the two paths is provided in the other part which is used jointly by the first and the second paths, the said means which are provided to ensure that the gain in the second path remains higher than the gain in the first path being adjusted in such a way that the electro-acoustic loop which is closed by the second path is permanently the source of unannoying oscillations of a weak amplitude a variation induced by the gain of the said further variable-gain amplifier determining a gain variation in the opposite direction in the other part.

19. A telephone set as claimed in Claim 18, charac-terized in that the variation in the gain of the said further amplifier is controlled by a detector detecting the speech signal generated by a microphone in the receiving part.