US2548668A - Electric signaling system - Google Patents

Description

April 10, 1951 B. M. HADFIELD ELECTRIC SIGNALING SYSTEM Filed July 8, 1946 FIG? INVENTOR BERTRAM MORTON HADFIELQ BY% dfl ATTORNEY Patented Apr. 10, 1951 ELECTRIC SIGNALING SYSTEM Bertram MortonHad'field, Harrow Weald, England, assignor to Automatic Electric Laboratories, Inc., Chicago, 111., a corporation of Delaware Application July 8, 1946, Serial No. 681,797 In Great Britain July 16; 1945" 4 Claims. 1

The present invention relates to circuit arrangements including athermionic valve and in its broadest aspect is particularly concerned with arrangements for biasing the control grid of the valve.

One of the objects of the invention is to provide improved arrangements for applying a biasing voltage to the control grid of a thermionic valve which is proportional to an input signal having positive and negative excursions.

According to the invention, the biasing voltage to the control grid of a thermionic valve pro.- portional to an input signal having. positive and negativeexcursions in which the biasing voltage is developed across a resistance in the cathode lead and the resistance is adapted to be alternate-. 1y decoupled and undecoupled by means of a switching device under the control of the input signal.

The invention is also concerned with electrical signalling systems and particularly with responding equipment for use in such systems.

A further object of the invention is to provide circuit arrangements including aresponding device which will respond at any given fractional value of an electrical signal which varies with time independently of slow variations in the absolute value of the signal and without taking any current from the input signal and without affecting the change of output current for a given change of input.

A particular object of the invention is to enable a relay or like device to respond faithfully on a time basis to input signals having an initial predetermined time duration and whose wavefronts have been distorted by the transmission and receiving equipment'which form an essential part of the system. Such signals may, for instance, be the pulse signals employed in automatic telephone dialling systems .or telegraph signals which are transmitted over a reactiveline or over frequency selective networks.

Generally speaking such received signals have successive wavefronts which are similarly distorted with time but are of opposing signs, and the operate and/or release of the relay or like device is then required to take place at a fractional value of one 'half the total change of the input signal; in other cases (e. g. the loop/disconnect method of transmission) the successive wavefronts may. not be similar and the desired fractional operation is greater or less than one half.

It is therefore another object of the invention to provide simple circuit arrangementsv whereby the same number of-componentsmay be readily adjusted in situto give the desired operation at any given fractional valueof the input and with any given change or tolerance in the components with sample or time.

A furtherobject-of the invention is toprovide arrangements whereby the-relay or like is held in the unoperated'condition should the input cease for an indefinite time.

According to one feature of the invention, in circuit arrangements adapted to respond to an electrical signal which varies withtime and including a thermionic valve having at least an anode, a cathode and a control grid with a responding device located in the anode circuit, in order to enable said device to respond at any given fractional value of the electrical signal which is applied to the grid/cathode circuit, a resistance in the cathode lead is alternately decoupled and undecoupled'by means of a switching device under the control of the electrical signal whereby the cathode current is adapted to execute positive and negative, excursions about a meanval-ue and the ratio of the current change in one di rection to the total current change is arranged to be substantially. independent of the electrical sig nal.

According to one feature of the invention, in circuit arrangements adapted to respond to an electrical signal which varies with time and including a thermionic valve having at least an anode, a cathode and a control grid with a responding device located in the anode circuit, in order to enable said device to respond at any. given fractional value of the electrical signal which isapplied to the grid/cathode circuit the cathode current is adapted to execute positive and negative excursions about a meanvalue and the ratio of the current change in one direction to the total current change is arranged'to besubstantially independent of=theelectrical signal due to the provision of a biasing network associated with the grid/cathode circuit and, including at least one asymmetrical conducting device.

According to a furtherfeature of. the invention, in circuit arrangements adapted to respond to an electrical signal which varies with time and in.- cluding a thermionic valve having, at least an anode, a cathode and a. control. grid with are! sponding device located in the anode circuit, in order to enable said device to, respond to any. given fractional value of the electrical signal which is-applied to the grid/cathode circuit, a biasing network is associatedjwith the grid/cathode circuit and includes a resistance in the cathode lead, the arrangement being such that the change of voltage in one direction across the resistance due to the change of cathode current on the application of the electrical signal to the grid/cathode circuit serves to apply a bias to the control grid proportional to the electrical signal to cause the cathode current to execute positive and negative excursions about a mean valueand to cause the ratio of the current change in one direction to the total current'change to be substantially entirely dependent on the value of the resistance.

According to another feature of the invention, in circuit arrangements adapted to respond to regularly recurrent impulses and including a thermionic valve having at least'an anode, a cathode and a control grid with a relay located in the anode circuit, in order to enable the operation of the relay to; take place at any given fractional value of the amplitude of each impulse a biasing network is provided comprising a cathode resistance in parallel with a condenser, the ends of the resistance and condenser remote from the cathode being connected through a metal rectifier so arranged that the cathode current executes positive and negative excursions about a mean value and the ratio of the current change in one direction to the total current change is determined substantially entirely by the value of the resistance.

In one embodiment of the invention the biasing network consists of a resistance in the cathode lead in parallel with a condenser, the ends of the resistance and condenser remote from the cathode being connected through a metal rectifier or a plurality of condensers may be employed and a metal rectifier is connected between the ends of each pair of condensers. The metal rectifiers are so poled as to be rendered conducting on an increase or a decrease of the cathode current according as to whether the responding device or relay is to operate on the positive or negative excursions respectively of the cathode current.

The invention will be better understood from the following description of a number of embodiments shown by way of example in the ac-' companying drawings comprising Figs. 1 to 7.

In Fig. l a triode valve V1 is shown with cathode resistance RI connected to the negative of battery BI, and an anode resistance R2 (simulating a relay or like device) connected to the positive of BI via earth. The series condenser CI and rectifier MRI are connected across RI, with the positive of MRI connected to the negative of BI. The input is applied via terminals l and 2 to the control grid of VI and the junction of CI and MRI negative. A schematic form of circuit producing typical input signals has been shown connected to the above simple form of the invention by dotted lines. A battery of voltage E is connected to the fixed contacts of a changeover switch A, the tongue being connected to a series resistance/condenser circuit, R, C, and the circuit completed so that as A operates the voltage waveform with time on C consists of exponential decays. and increments between the limits E and zero. The components R, C, have therefore introduced distorting wavefronts to the initial rectilinear time pulses from the contact A; in this casev the successive wavefronts have similar form. The signal therefore consists of the change of voltage on C, that is, E. It is desired that the relay shall operate and release at around successive input voltages of E/Z, since the input A contact time interval is only reproduced at these voltages on the wavefronts (or more strictly, at voltages whose arithmetic mean is E/ 2) When the input has been zero for a sufiicient time the voltage on Ci equals that on RI (since the back resistance of MRI is not infinite), and a steady current flows through the valve due to the self bias on RI. Application of input signal voltage variations will modulate this current, so that it forms the reference current for the valve circuit, and in particular the relay. The object of the circuit is to ensure that the excursions ofthese modulations are always the same fraction of the total signal current variation irrespective of the magnitude of the signal variation.

Let the mutual resistance of the valve circuit be i g i. e. the changeof grid/cathode voltage divided by the change of cathode current, when VI, RI and R2 are all in series with the battery BI. Then for a steady positive input of E as shown, the change of cathode current will be E/(R+Eg) where R is the value of RI, because rectifier MRl freely conducts for an increment of voltage on R. If now E falls to zero, and during this time the voltage on CI can be considered as constant (owing to the high backward resistance of MRI), then the whole of the signal input change of E is applied to the grid/cathode, and the resulting fall of cathode current is E/Rg. Hence the ratio of positive current change to total current change is Rg/(l m-R), which is independent of the input signal magnitude E. Put in another way, the positive increment of voltage on RI due to E is Rg/(y+R) times E, and acts as a bias on the subsequent total change of input E, thus causing the signal change of current to have positive and negative excursions about the steady value, and since the bias is proportional to E the fractional values of these excursions remain constant with changes in the magnitude of the signal.

Since Rg is comprised by the sum of the mutual grid resistance of the valve alone, R9 (1. e. the reciprocal of its slope), the anode resistance divided by the amplification factor, ,u, (i. e. RZ/a), and the cathode resistance divided by ft (i. e. EI/ l, it is clear that the determining ratio Rg/(Rg-l-R) can be adjusted at will to have any desired value between zero and unity, by variation of R3, in the case of the waveform cited in Fig. 1 having similar successive wavefronts, operation at a fractional value of onehalf the total signal current excursion will be obtained when R/(R g-I-R) is made'one-half. It is possible therefore, not only to accommodate different fractional values but also changes of the mutual resistance, Rg, of the valve with sample or life, by making the cathode resistance variable to any required degree (as denoted in Figs. 1 and 2 by the dotted arrow associated with resistances RI and R3 respectively). With this simple arrangement such adjustment will also alter the normal steady non-signal cathode current and means would have to be provided to allow for this in practice; for example, where R2 is a relay, the current through a biasing winding would have to be altered correspondingly. Other arrangements will be described later which; will avoid this necessity.

A simple method of arriving at the given fractional operating value, which is generally aping when the control grid is taken successively to (1) that end of the cathode resistance remote from the cathode and to which the first rectifier is connected, and (2) that end of RI nearest the cathode and to which the first condenser is joined. These currents then have a ratio equal to Rg/(R-l-Rg), which is as described before, the ratio of the positive signal current change to the total signal current change.

The mutual grid resistance of the valve, R'g has so far been assumed implicitly to be constant, whereas as is well known it varies probably as the reciprocal of the one-third power of the current. That is to say, over the higher current range of the valve, the variation of Rg is small. Now it is over this range that the bias on RI is produced by the maximum value of the signal, and moreover R! then produces negative feedback. Hence, the determination of the correct operating conditions by RI is performed under almost linear conditions, so that the assumption that By is constant is correct for all practical purposes. The operating point will be determined to a close degree of accuracy by the above theory and adjustment, and any final adjustment can be done in situ when the desired signal is applied. The maximum positive and negative excursions of current will not be properly related (owing to the curvature of the valve characteristic at small currents) but this is in general of no significance provided the relay or other device operates and releases at the correct time intervals, and provided of course that such excursions do give enough energy into the device for the purposes contemplated.

It will be observed that provided the normal current range of the valve is used (i. e. the maximum limited to a value at which no grid current passes), then no current is drawn from the input signal, other than that due to the inter-electrode capacities. Also the total signal current change obtainable is the same as would have been obtained had the valve been used as a normal amplifier, e xcept that the mutual resistance of the circuit Rg has been increased slightly from normal by the amount R/ Taking as an example the case when R equals R9, the increase in the mutual resistance of the circuit to the total signal change will be normally 10% at the most. When account is taken of the reduction of battery voltage in the normal circuit by decoupling RI, it will be seen that total change of signal current in the present invention is practically speaking the same as heretofore, so that the invention may be readily applied to existing valve/relay circuits without loss of sensitivity.

It will be appreciated that the success of the invention depends largely on the rapid charge sary to sustain to this degree up to the time when the signal regains the desired fractional amplitude. With the simple circuit shown in Fig. 1 this means that the backward time constant of CI and MRI must be some 20 times the duration of the longest signal, since the discharge of CI is exponential. While this may be secured by proper selection of the values of CI 6 and MRI, it imposes a somewhat lengthy period of readjustment on the circuit should the maximum value of the signal be reduced for some reason outside the control of the receivin end.

There is bound to be some distortion due to this cause whatever the readjustment time of the circuit, but it is possible to reduce the number of signal pulses which are so distorted by the following modification to the biasing circuit.

The arrangement of Fig. 2 shows the basic elements of Fig. l with the addition of a further condenser/rectifier circuit, 03, MR3. If these additional elements be designed correctly (e. g. so that they do not draw a significant current from C2), then the initial rate of discharge is zero, is attained relatively slower therefore, and any lower value relatively quicker. It is possible in fact to reduce the readjustment time of the circuit by about 10 times with this circuit, for the same sustaining time to 95%. It is clear of course, that should this not be a sufiicient reduction, then further condenser/- rectifier circuits can be added.

In practice the rectifier will impart a voltage to its condenser, which is to all intents and purposes constant and independent of the source voltage on the cathode resistance. The voltage is that required just to cause the rectifier to conduct, and will be apparent after the first signal pulse, thus tending to cause first-pulse distortion. This may be avoided by shunting the condenser with a resistance low in comparison with the backward resistance of the rectifier, so that a priming current is normally drawn through the rectifier. The condenser/resistance then becomes the determinin factor in the circuit time constant, and has the advantage that the performance of the circuit is rendered largely independent of initial or subsequent values of the rectifier back resistance.

If the rectifiers MR2, MR3 be connected in the reverse sense to that shown in Fig. 2, and the applied signal be also reversed in sign (i. e. vary between a negative Voltage on the control grid and zero), then the circuit of the invention will perform in the same manner as before. The change of voltage stored on the condensers during the maximum negative value of the signal will bias the subsequent change of cathode current when the signal attains zero voltages, the discharge of the condensers during this regime being slow.

If the input signal consist of rectifier alternating current or of alternating current, then the invention will work in a similar fashion, since the condenser will charge to the peak value of the input (either the positive or negative according to the connection of the rectifiers) When the input signal envelope reduces in magnitude to zero, owing to the signal pulse, the envelope of the signal cathode current will also fall through the steady state cathode current value to a value less than the steady cathode current. By consideration of the mean value of the cathode current over this time, the resistance R may be computed in terms of Rg to enable the relay to operate and release at the appropriate time intervals on the envelope.

If the input signal be biased in any way such that it may be analysed into a signal change plus a constant independent of time, then the conditions for the invention to work are precisely the same as regards the signal change (i. 'e. the value of R! in terms of R9), but the steady state cathode current forming the reference current for the 7 relay will be correspondingly altered. This is clear from the foregoing description, because such steady state cathode current as exists therein is solely due to the battery B, which can be regarded if one chooses as a bias on the signal.

Other methods will now be described whereby the fractional operate and release values of the circuit may be adjusted for differing wavefronts and/or differing circuit component values.

Fig. 3 shows the essential parts of the circuit as for Fig. l but with the cathode resistance R now fixed in value and having the rectifier connected to a tap on the resistance. The ratio of positive current excursion from the steady value to the total subsequent signal current change is now Eg/(Rg-i-pR) where pR is that portion of R5 between the tap and the cathode. For given values of R5 and By the fractional value can be determined in the terms of p, and hence for a given waveform the potentiometer R5 can be calibrated in terms of Ry, so that differing valve resistances can be used. This modification however still suffers from the defect that the steady reference current of the circuit alters as p is altered.

Fig. 4 shows the same essential parts as Fig. 1 but with the condenser now connected to a tap on a fixed cathode resistance RI. The ratio of positive current excursion from the steady value to the total subsequent signal current change is now (q.R+1 2g) (R+1 g) where R is the value of R1 and q.R is that portion of RI between the tap and the cathode. Again for given values of RT and fig the fractional value can be determined in terms of q and for a given waveform the potentiometer R? can be calibrated in terms of R9, so that differing valve resistances can be used. In particular both Figs. 3 and 4 can cope with any valve having a mutual resistance greater than that designed for, with the one design of potentiometer. The arrangement of Fig. 4 however, has the practical advantage that variation of q does not alter the steady value of current,

since this is dependent on (R+Rg) so that for a given valve q may be adjusted at will to meet differing wavefronts. Also the adjustment of q for a specific fractional value based on the method of joining the grid successively to the negative of B5 and to the tap, does not now require repetition when q is adjusted, since such adjustment does not alter the first cathode current reading. Whilst the above method does permit the fractional operate or release value of the circuit to be adjusted without altering the steady circuit current for a given value, the current will alter as between different samples of a valve or with life. If it is desired to keep the operating conditions of the relay constant, 1. e. the steady circuit current constant, then the following methods may be used.

With the circuit of Fig. 4, in which R? is fixed in value and the fractional operate or release value is adjusted by a tap on the resistance, the only remaining variables causing differences of steady cathode current are the effective amplification factor and mutual grid resistance of the valve. Since the Worst valve of a given class will have to be tolerated, it is reasonable to base the design of the relay or like device on this valve, and to provide means whereby all valves are brought to this state as regards the circuit currents and voltages. Hence means for reducing the mutual resistance of the valves to such a value as will give the same cathode circuit our- 8 rent, combined with any of the described means for obtaining the required fractional operate or release value, will accomplish the desired effect.

The arrangement of Fig. 5 shows a triode valve V5 and the essential parts of the circuit. The components C6, RH and MR6 correspond to C5, R! and MR5 of Fig. 4 and determine the ratio of the positive excursion to the total current change. Two alternatives for the adjustment of the steady current are shown; one a variable resistance R9 in the anode lead, and second a variable resistance R in the cathode lead. Either method is satisfactory, the difference lying in the respective magnitudes of R9 and RI 0 for a given range of valves. Thus R9 will need to cover approximately ,u. times the range of RIO. The cathode resistance method is to be preferred on the whole, since it also contributes negative feedback to the valve itself and makes the assumption that R9 is linear even more valid. Where the valve V5 is a pentode or tetrode valve, which means the interposition of a screen electrode between the control grid and the anode connected to a convenient positive battery, say, earth of B5, the correct position for R9 is in the screen lead if this method is used as shown in Fig. 7, Where RIB is the resistance in question.

There are, of course, other ways of obtaining the same cathod current from valve to valve, such as a variable control grid bias, or a variable anode or screen potential, but the practical application of these methods suffers from defects obvious to those skilled in the art.

It has been assumed so far that the only effect produced by the relay on the circuit is due to its direct current resistance, in that the muwhole, however, to use a pentode or tetrode valve,v

in which whilst the relay still needs such treatment, more scope is allowable for transient signal anode voltage changes without affecting the cathode current; in addition, no account need be taken of even the direct current resistance of the relay in computing the mutual grid resistance fig of the circuit.

Fig. 6 shows a form of the circuit of the invention using a tetrode valve V6. The relay X in the anode lead is chosen to have a direct current resistance such that the minimum signal anode voltage exceeds the well-known knee value, and hence does not affect the cathode current. The screen is connected for convenience to the positive of B8 via earth, whilst the functions of Hi3, Cl and MRI are as described before. Relay X may be shunted by a condenser C9 to reduce the transient signal voltage on the relay and/or to provide an elementary low pass filter effect in conjunction with the relay series inductance and resistance so that unwanted low frequency ripple on the signal envelope may be reduced to negligible proportions in the relay current, thus avoiding random distortion of the relay pulses. In this case of unwanted ripple on the signal, the condenser may also be connected so as to include R14 as shown by C8, thus reducing this ripple on RM as well,

so that the circuit works as designed to th mean signal envelope change with time.

Fig. 6 also shows an additional feature of the invention, concerned with ensuring that the relay is held in the released condition when the signal disappears for a long time, i. e. when the voltage on CT falls to its steady state value. The circuit addition consists of a rectifier MR8 connected with its positive pole to the junction of Rl3 and RM and its negative pole to the junction of resistances RI5 and RIG placed in series across a convenient battery such as B6.

It is arranged that the potential across RIG just exceeds that'across RM when the steady cathode current conditions exist, so that MR8 is then conducting and the resulting additional current flow down Rl l biases the valve to give a current less than normal, so holding the relay non-operated. When a signal is applied to the control grid'such as will normally produce a voltage on RIG exceeding that on RIG, then the rectifier ceases conducting and the circuit works normally as regards the generation of the correct additional bias on C1 to ensure correct fractional operation and release of the relay. During the signal pulse RM is shunted by RIB, so altering the circuit mutual grid resistance R g by the difference between RM/ and but the difference is small if the circuit is correctly designed to avoid a large current drain on B6 due to R15 and RIB.

This additional circuit, or others of similar action, may be applied to any of the previous figures, but will be the more easily applied the more the fundamental circuit ensures a constant non-signal input voltage across RN or that portion of Rid to which it is applied.

In designing the circuit of the invention, for instance that of Fig. 6, it is arranged that for a valve with the maximum value of Rg and the minimum current (including the desired life effects) Rl3 shall be zero, when the value of RM and the degree of tapping variations are easily found for the desired wavefronts and fractional distortionless operation of the relay as described before. An upper limit is then fixed for Ri3 by specifying the minimum value for R9. The steady cathode current for all valves used in the circuit will then be known, and from the normal steady voltage drop on RM, the potential divider R15, RIB, may be calculated for a given Value of release ampere-turns in the relay. In setting -up the circuit, .the input terminals are shorted or the signal is reduced to its minimum value, and Rl3 is adjusted to obtain the desired steady cathode current with MR! disconnected. The desired fractional operate and release value is then obtained experimentally by applying the signal waveform and adjusting the tap on RM, or by transferring the control gridto the tap and adjusting this to obtain a cathode current which divided into the steady current gives the desired fraction. In practice this tap should not have to vary by more than from unity to 0.8 from the oathode end of RM. With that valve and with life the only readjustment normally necessary will be to reduce Rl3 to re-attain the steady normal cathode current, and when Rl3 is reduced to zero an automatic indication is given that the valve needs replacing.

released when the signal ceases fora long time,

can be secured without-the additional non-linear bias circuit described, by usingapolarised relay of the type in which a sufiicient magnet toggle action exists, and having a bias winding whose ampere turns are equaland opposite to those produced by the steady anode current.

The circuit may if necessary be arranged to compensate for thetransittime of the relay, as regards the output on a given contact, and with changes in input signal level, by altering the fractionaloperate and release value of the circuit by an amount appropriate to the transit time and the signal wavefront velocity, so that the output is rendered-completely distortionless at all the designlevels.

I claim:

1. In a signalling system, an electric discharge valve having a cathode, a control grid and an anode, a source of potential connected to said anode to provide a voltage tosaid anode toj'enable said valve to conduct, a control circuit connected to said control grid for iprovidingan input signal to said valve, another source of potential, a control contact included within said control circuit and having a first and a second position, a circuit connecting said control circuit to said other source of potential in case said control contact is in the first position forenergizing said control circuit, said control circuit deenergizing in case said control contact is in the second position, a cathode resistor connected to said cathode and said first source of potential to provide a biasing potential for said valve in response to said valve conducting by the application of a potential of a predetermined value from said control circuit to said control grid, a cathode condenser and an asymmetric conductor serially connected to said cathode and said source of potential for providing a biasing network for said valve in conjunction with said cathode resistor, a conductor interconnecting said biasing network, said control circuit and said other source of potential for providing a common connection, said cathode condenser and said asymmetric conductor being connected in multiple with said cathode resistor for charging said cathode condenser to a value corresponding to the potential across said cathode resistor, the polarity of said asymmetric conductor being effective to cause substantially all of the charge on said cathode condenser to be retained in case the potential across said cathode resistor decreases to maintain a fixed bias on said valve at a polarity opposing the conductivity of said valve.

2. A signalling system as claimed in claim 1 and including a means connecting said anode and said first source of potential and operated in response to the conduction of said valve each time said control contact changes position for providing an output signal in unison with the changing of position of said control contact.

3. In a signalling system, an electric discharge valve having a cathode, a control grid and an anode, a source of potential connected to said anode to provide a voltage to said anode to enable said valve to conduct, a resistance-capacitance circuit connected to said control grid for providing a control circuit to said valve, another source of potential, a control contact having a first and second position, a circuit including said control contact in the first position connecting said resistance-capacitance circuit to said other source of potential for charging said The facility of ensuring that the relay is held resistance-capacitance circuit, a second circuit 11 for discharging said resistance-capacitance circuit in case said control contact is in the second position, a cathode resistor connected to said cathode and said first source of potential to provide a biasing potential for said valve in response to said valve conducting by the application of a charge of a predetermined value from said resistance-capacitance circuit to said control grid, a cathode condenser and a rectifier serially connected to said cathode and said first source of potential for providing a biasing network for said valve in conjunction with said cathode resistor, a conductor interconnecting said biasing network, said resistance-capacitance network and said other source of potential for providing a common connection, said cathode condenser and said rectifier being connected in multiple with said cathode resistor for charging said cathode condenser to a value corresponding to the potential across said cathode resistor, the polarity of said rectifier being effective to cause substantially all of the charge on said cathode condenser to be retained in case the potential across said cathode resistor decreases to maintain a fixed bias on said valve at a polarity opposing the conductivity of said valve.

4. A signalling system as claimed in claim 3 including a means connecting said anode and said first source of potential and operated in response to the conduction of said valve each time said control contact changes position for providing an output signal in unison with the changing of position of said control contact.

BERTRAM MORTON HADFIELD.

REFERENCES CITED The following references are of record in the file of this patent:

UNITED STATES PATENTS Number Name Date 2,282,182 Gulliksen May 5, 1942 2,360,721 Rose Oct. 1'7, 1944 2,361,172 Brown Oct. 22, 1944 2,363,753 Smith Nov. 28, 1944 2,366,060 Schneider Dec. 26, 1944 2,372,516 Rechton Mar 2'7, 1945

Images