US2313098A - Method and means for reproduction of sound frequency vibrations - Google Patents

Description

March 9, 1943. F. H. SHEPARD-JR 2,313,093

METHOD AND MEANS FOR REPRODUCTION OF SOUND FREQUENCY VIBRATIONS Filed April 28, 1942 3 Sheets-Sheet l 0U TPU T IN VEN TOR. Francis H. ShefmraJn HTTORNEYS Mardl 9, 1943. SHEPARD, JR 2,313,098

METHOD AND MEANS FOR REPRODUCTION OF SOUNDFREQUENCY VIBRATIONS Filed A ril 28, 1942 a Sheets-Sheet 2 1+5 Dz i %.25MEG. 10" hasq. N l

nu nun I 3:32;; 1 zoo'om 150 as i 42 I B INVENTOR.

BY v

H TTORNE Y5 Patented Mar. 9, 1943 METHOD AND MEANS FOR REPRODUCTION OF SOUND FREQUENCY VIBRATIONS Francis H. Shepard, Jr., Merchantville, N. J.

Application April 28, 1942, Serial No. 440,786

12 Claims.

This invention relates to amplifiers, and more particularly to new and improved methods for the amplification or reproduction of sound frequency vibrations.

This application is a continuation in part of m copending application Serial No. 330,056,

filed April 17, 1940, and is also related to my copending application Serial No. 359,856, filed October 5, 1940, for a System for compensating anode supply potential variations.

The present practice, when amplifying sounds, vibrations or impulses received by a microphone or other pick-up device, is to keep the amplification as distortlonless as practicable. It is generally assumed that any distortion introduced by the amplifier is undesirable and great efforts are made to keep this distortion as low as possible. When the range of sound intensity that is desired to amplify is great, it is customary to monitor the gain either manually or automatically in such a manner that the distortion of the amplifier is kept as low as possible. Care is exercised to limit the rate of change in gain to avoid undesirable effects well known to those familiar with the art.

It is well known to those versed in the hearing art that a distortion is introduced into the sound signals somewhere between the eardrum and the brain. Curves showing this characteristic are shown in Hearing" by Stevens and Davis, page 195, Fig. 82, published by John Wiley & Son, Inc., New York. This characteristic becomes practically a straight line within certain limits when plotted on logarithmic paper. Over the straight portion of this characteristic the curve can be approximately expressed by the equation Pe=V" where P; is sound pressure picked up by the ear,

.V is the voltage or'stimulus to the brain, and n is an arbitrary exponent introduced by the hearing mechanism of the ear and generally greater than 1. It should be noted here that the exponent n in the human ear is different for diiierent frequencies, 1. e.. different wavefront steepnesses. In general in the human ear the exponent is greatest at frequencies in the order of 3000 cycles.

It is also generally recognized that the human ear has some sort of compensatory mechanism comparable to that of the eye, that causes the ear to be more sensitive in a quiet room than when in noisy surroundings. This accommodation of the ear takes place over a relatively long period of time compared to the rate of change of sound intensities. It may take a considerable time for the ear to reach its final state of accommodation for any one sound intensity level. The

accommodation takes the form of a change of slope of a straight line plotted on the logarithmic scale, that is, a change which is roughly represented by a change in the value of the exponents n. The characteristic remains approximately straight over roughly the same range of V stimulus to the brain for all values n. It is also recognized in medical science that the sound sensitive mechanism of the ear is not directlyv sensitive to certain low frequencies, but that it responds to harmonics thereof. Accordingly the presence of a frequency can be recognized by the brain even though the fundamental may be suppressed.

From the above discussion it can be seen that the brain is accustomed to interpreting a distorted wave or a stimulus and that it has trained itself to disregard or hear as natural this type of distortion. It should also be noted that the range of stimuli intensites received by the brain is considerably less than the range of sound pressures actually listened to.

An object of my invention is to prediztort the waves representing sound vibrations in a manner similar to the distortions produced in the human car. This may be accomplished by distorting the instantaneous amplitude of these waves to obtain a compression or expansion of the sound volume and a change of wave shape in a form similar to that produced directly in the human ear. The characteristics of the distorting means. preferably approximate, Vm=P where Vm is the sound pressure picked up by the microphone or other pick-up device, such as the audio detector of a radio receiver, Fe is the sound pressure fed to the ear and q is an arbitrary exponent introduced by the distorting means.

The characteristic from the distorting device to the brain is practically expressed by Vm=P" where nq is a new overall exponent. As far as the brain is concerned, it apparently thinks that the new exponent is a result of an accommodation of the ear itself and so it is not conscious of the distortions introduced. Also, even though the range of sound intensities actually fed to the ear is considerably less than the range of sound intensities picked up by the pick-up device, the brain still feels or senses the original dynamic range. It should be noted here that the characteristics discussed are instantaneous and involve no time delay as in the conventional known systems of volume contraction and expansion, such as is shown, for example, in the Case Patent No. 2,072,708.

In general this distortion effect is obtained by using a non-linear transducing device or combination of devices. In cases wherein the natural curvature of the device is not sufficiently great or suitable it can be modified by raising its first derivative to a power and utilizing the integral function of this resultant. This may be accomplished by utilizing an amplifier which itself may have a non-linear characteristic or may be operated in conjunction with other apparatus having non-linear characteristics. The non-linear characteristic in all cases should have a peaked derivative. It should be understood that this characteristic refers to the overall characteristic of the system and may be obtained by utilizing a combination of the various elements for its derivation.

My invention also has further applications making recordings on any medium or in transmitting signals over telephone lines, radio links, or other medium where the limit to the dynamic range of signals it is possible to handle is controlled by the noise level on one hand and by the overloading, over-cutting or over-modulation and the like on the same medium on the other hand. As above explained, if the proper type of distortion is introduced on these various types of apparatus, the range of signal intensities will be considerably less than the original range of wave intensities. However, as explained above when listening to sound the brain will not be conscious of the degree of dynamic range reduc tion that has taken place. Thus not only will the sensation of dynamic balance be present, but because of the frequency response characteristics of the hearing mechanism the low frequency components will also be sensed even though the actual low frequencies may not be radiated from the speaker.

In my invention a further object thereof comprises an arrangement to restore completely or in part the original character of the wave by making the play-back or reproducing system have the characteristics to correct for the distortion previously introduced, namely, the approximate characteristic where R is the signal level on the recording medium and Pa and q have the same significance as previously recited. It should be noted that even though the value of q in the first part of the system dififers from the value of q in the reproducing system, no harm is done because the resultant overall effect is Vm=P in which q will differ from unity and hence as explained above the type of distortion that is introduced will be unnoticed by the brain of the listener.

While I have discussed above certain objects of my invention in the manner in which it will operate, a clear understanding thereof may be obtained from the particular description of a few preferred embodiments thereof made in connection with the accompanying drawings, in which- Figure 1 shows explanatory curves used to describe the operation of the device;

Figure 2 is a circuit diagram illustrating one manner of practicing my invention;

Figure 3 is a circuit diagram of a modification of Figure 2 with the addition thereof of means for compensating anode supply potential variations known as hum bucking, as disclosed in my copending application Serial No. 359,856 above referred to;

Figure 4 is another modification showing a different arrangement for hum bucking, including a degenerative feedback to the screen of the first audio stage;

Figure 5 is another modification showing still another arrangement for hum bucking and degeneration with means for shorting out positive and negative feedback if desired.

Figure 6 is another modification showing still another arrangement with positive feedback derived from the plate circuit.

The circuit of Figure 2 is identical with the circuit of Figure 11 of my application Serial No. 330,056 above referred to. It shows a partial circuit diagram of a radio receiver indicating a tube 2| which may be the second detector of a superheterodyne set or a detector of a tuned radio frequency set. The output of tube 2| may be applied to an audio amplifier circuit through a manual volume control indicated at 2 through the network consisting of condenser 3 and resistance 4. An amplifier tube 90, which may be a screen grid tube as shown, amplifies the signal and the output thereof is impressed upon a second amplitlertube 9| across a network consisting of a condenser and resistance 8. The tube 9| which may be a tetrode, as shown, further amplifies the signals and the output thereof may be supplied to the loudspeaker. In the cathode lead of tube 9| is a biasing resistance 5 which is tapped at Hill to provide a feedback path through the reslstance III across condenser I02 to the cathode circuit of tube 90. The tubes and ii are operated so as to have an s-shaped characteristic curve. It will be noted that the resistance coupling network between the tubes consisting of condenser l and resistance 8 will have their values chosen so as to reduce the passage of frequencies below, say, 200 cycles. The feedback circuit consisting of resistance llll and condenser I02 will supply feedback potential from the cathode circuit of tube 9| to the cathode circuit of tube 90. which will be in such phase as to be regenerative. This network will attenuate signals of a frequency above, say, cycles. The result of the operation of the tube resistance networks I, 8 and ii and I02, will be to cause the greatest amount of regeneration at, say, 80'cycles, a frequency well below that which will be efllciently radiated by the loudspeaker. The result will be that the S-shaped characteristic is modified to have a steeper slope near its center for a range of frequencies on both sides of the 80 cycles just referred to, with the result that odd harmonics of this range of frequencies will be generated and accentuated so that they can be radiated by the loudspeaker and will make the listener believe that he is hearing a low note of the original frequency, as has been explained above. Also low frequency waves which are of high enough frequency to be made audible by radiation from the loudspeaker will be enormously amplified to overcome the deficiency of the normal human ear in responding. to these low frequencies and to overcome other characteristics of radio sets which inherently tend to discriminate against low frequencies. This gives the illusion that a given radio set has a greatly increased power handling capability.

This circuit just described is by itself effective as a means of damping speaker resonance, because speaker resonance results in a reduction of current in the speaker transformer. This reduction of A. C. plate current reflects itself into the cathode circuit causing a reduction of cathode current which in turn causes a reduction of positive feedback, which in turn reduces the effective gain of the amplifier at the point where the efficiency of the speaker has increased due to its resonance. Since, however, variations of plate current do not cause as great variations in cathode current because of the characteristic of the high impedance output tube, in which current variations of the screen partially balance those of the plate, a negative feedback is introduced to cause a reduction of efiective impedance of the output tube and to cause a greater portion I 2,318,098 of current variations occurring in the output tube to appear in the cathode or feedback part of the circuit. This is accomplished by means of the resistance 9 connected from the plate of the output tube 8| to the plate of the driver tube 80. .It will be understood that other conventional methods of negative feedback may also be used, such as coupling from the voice coil or any secondary winding of the output transformer back into a suitable part of the drive circuit, as illustrated in Figures 4 and 5. The point in connection with this auxiliary speaker damping circuit in whatever form it may be used, is that it cooperates effectively with the regenerative circuit which is used for other purposes, to provide effective speaker damping. I

At the point of resonance in the output, if

' constant current is fed to the output, as is essentially the case for a tube having an infinite output impedance, the voltage across the load will rise. There will be little resultant change in current. However, if by the use of suitable negative feedback the effective plate impedance of the driving tube is lowered, the change in voltage across the speaker due to its impedance change at resonance, will cause a substantial change in current, which will appear in the part of the circuit from which positive feedback is derived.

In Figure 3 I have shown another circuit diagram illustrating a modification of my invention in which, aswell as in succeeding figures, corresponding parts are correspondingly numbered to those in Figure 2 In this diagram the regenerative feedback circuit is from cathode to cathode through resistances Hll, IOIA, and IUIB, across condensers I02 and HA. Such a double section filter will give sharper frequency discrimination than the simple filter shown in Figure 2.

In Figure 3 I show a hum bucking condenser I connected from the screen grid of tube 90 to 3+. This operates to apply to the screen of tube 90 any hum voltages which may exist in the B supply and which accordingly would be applied to the anodes of the tubes. This condenser, in conjunction with condenser ll, acts as a voltage divider to apply a suitable portion of the variational voltages that may appear in the B supply to the screen in the proper phase and magnitude to cause plate current variations of such phase and magnitude a to oppose hum voltages that may be fed through the plate resistor of the driver, or to the plate and screen of the output tube. This has been explained in my copending application Serial No. 359,856 above referred to. It can be seen that hum bucked in this manner causes no extraneous hum voltages or currents to appear in the positive feedback parts of the circuit, and therefore these parts may be freely used to provide the regenerative feedback which is necessary for the production of th synthetic low notes without any hindrance from any hum which may be present and without any danger of unduly regeneratin any such hum and thereby making it highly objectionable. Such a hum bucking scheme is therefore particularly advantageous in connection with my distorting circuit shown. This circuit is also effective not only in bucking the fundamental hum frequency but all harmonics thereof. Many of the conventional hum bucking arrangements which utilize transformers, phase shifting devices, etc. are not as effective in this respect.

The auxiliary resistance 8 could also, of course,

be used in this circuit for more efi'ective damping of speaker resonance, if necessary or desirable, with the same advantages when used in combination with my regenerative distorting circuit 'which have been pointed out in connection with Figure 2.

In Figure 4 I have shown another circuit variation. Figure 4 is essentially the same as the circuit shown in Figure 2, except that the voice coil winding is interposed between the by-pass condenser H from the screen of the driver tube to B This introduces negative feedback from the output into the driver circuit. Since the conventional by-pass cathode condenser I2 is shown across the bias resistor 5 for the output tube, the high frequency variational currents in the cathode circuit of the output tube are bypassed directly from cathode to ground. Hence only low frequency signal voltages will appear on the cathode of the output tube. These voltages are fed back through the non-frequency discriminatory resistor IOI to the cathode of the driver tube. (This can be used in conjunction with the cathode resistor 5 and the cathode impedance of the output tube as the feedback frequency discriminatory network. Resistor Ill acts as a non-frequency discriminatory feedback link from the cathodeof the output tube to the cathode of the driver.)

As explained above, this circuit, because of the negative feedback in combination with the positive feedback is extremely effective in controlling speaker resonance, and to a smaller degree, cabinet resonances.

In Figur 5 I have shown a circuit diagram of another modification of my invention in which the negative feedback is introduced from the voice coil to the cathode of the driver through a part of the positive feedback in order to lower the plate impedance of the output tube and to aid in damping speaker and cabinet resonances. The circuit otherwise operates as that shown in Figures 3 and 4. I have also shown in Figure 5 a switch 13 which can be used to short out the positive and negative feedback voltages so that the circuit will operate as a normal amplifier. This switch also acts to shortcut the unbypassed driver cathode resistance i8, and hence the degenerative losses that occur in this resistor. Thus the full gain of the driver may be realized. This is sometimes desirable when the radio is turned to extremely weak signals which are partially submerged in the noise level, or in static, thus making it unnecessary to have good fidelity. In other words, where it is desired to listen to a program on a particular station which is very weak or coming in very poorly, and the desire of the listener is to get that program regardless of fidelity he may close the switch l3 and thus take full advantage of the high gain of a pen.- tode driver although he will lose. the effect of fidelity which is otherwise produced by the dis tortion generated in my system.

Figure 6 shows an embodiment of my invention wherein voltage for positive feedback is obtained from the drop across a resistor M in series with the plate circuit of the output tube. This voltage is fed through the resistor l5 across condenser l6 and through the condenser H to the screen of the driver.

Condenser II, in combination with condenser Negative feedback for damping speaker resonance 'is introduced in the same manner as in Figure 4, wherein feedback voltage is fed to the screen of the driver. In other respects the operation of the circuit of Figure 6 is the same as the above-described circuits. This particular type of circuit is useful in connection with filamentary type tubes wherein cathode potentials are fixed.

The operation of my invention may be further described in connection with Figure 1. It has been previously stated that the network 1, 8 may have the value of its parameters so chosen as to reduce the passage of waves of a frequency below say 200 cycles, and that the feedback network I I02 of Figures 2 to 5, or [5, l6, H of Figure 6, may have the values of its parameters chosen to reduce the passage of waves of a frequency above say 100 cycles, and that the result of operation of the two networks will be to cause the greatest amount of regeneration at some selected low frequency, say 80 cycles, a frequency well below that which will b efficiently radiated by the loudspeaker.

Considering now that 80 cycles is the frequency of maximum regeneration, let us consider what will happen to waves of various frequencies and amplitudes.

A wave of high audio frequency, say 4000 cycles, will be amplified in the tube 90 and passed on to the tube 9| with substantially no attenuation. It will, however, be highly attenuated in the feedback circuit so that substantially no voltage of this frequency will be fed back to the tube 90. It will, however, be further amplified in the tube 9| and passed on to the loudspeaker. Since there is substantially no regeneration of a wave of this frequency it may be seen that the wave will be operating upon a composite characteristic substantially like that of curve 2 of Figure 1, being amplified but not being distorted a great deal as to wave form, such distortion as there is not producing any undesirable effects. It will further be noted that such a high frequency wave of low amplitude (low signal level) will be amplified more in proportion than a wave of high amplitude of the same frequency, so that greater amplification is automatically provided when needed for audibility.

Let us consider now what happens to a low frequency audio wave, say one of 80 cycles. This wave will be amplified in the tube 90, passed on somewhat attenuated, to the tube BI and again amplified. Then it will be selectively regenerated and again amplified in the tubes 90 and SI. Thus it will be seen that this wave will be regeneratively amplified, so that it may be considered to be operating on a characteristic similar to that of curve 4 of Figure 1. Also it will be symmetrically distorted so that its steepness is increased close to and on both sides of the zero axis, thus generating odd harmonics of the wave, particularly, of course, the lower odd harmonics. If this low frequency wave we are now considering is of high amplitude to begin with (high signal level) it will, of course, not be amplified in the same proportion in my system as a low frequency low amplitude wave. In fact by reference to the curves 2 and 4 of Figure 1 it will be noted that if the low frequency wave is of sufiicently high signal level it may be even amplifie less than a high frequency wave, as it will be noted that the curves 2 and I cross over at their extremities.

Thus in my system it will be noted that the following results are occurring:

1. A high frequency wave will be amplified only slightly more, in proportion, if of low signal level than if of high signal level.

2. A low frequency wave of low signal level will be enormously amplified.-.

3. A low frequency wave of high signal level will be amplified, but no more than, and if of a high enough level even less than, a high frequency wave.

4. The low frequency waves will be sufficiently distorted top'roduce odd harmonics, particularly the third and fifth harmonics, although there will be less of the fifth harmonic than of the third.

It is to be noted, 01' course, that although the 80 cycle wave is favored in its amplification, there will be a range of waves about this frequency which will be favored, although to a lesser extent.

It is also to be noted that all of these results are occurring simultaneously, and instantaneously, that is with no time delay. Thus waves of a frequency which may be high enough actually to be heard directly through the loudspeaker used, say waves of cycles, will be amplified sufficiently to be heard, even though of low signal level, it being well known that the human ear requires a much greater amplification of these low frequency waves than of waves of about 3000 cycles in order to be heard. My circuit may be used, however, with a loudspeaker which is incapable of reproducing waves as low as 80 cycles. As to these waves, the odd harmonics referred to will be produced and the hearer will have the illusion of hearing signals of this low frequency even though the loudspeaker is incapable of reproducing this low frequency itself. This result also occurs simultaneously with the others mentioned.

My circuit also simultaneously operates to keep anode supply potential variations out of the output circuit and to do this at the same time that the regenerative circuit is operating to secure the results mentioned above. At the same time it is also utilizing a cooperative relationship between the negative feedback and the positive feedback to damp speaker resonance.

My invention is capable of various modifications in addition to those disclosed herein. It is therefore to be limited only by the scope of the appended claims.

What is claimed is:

1. A system for conveying waves representing sound frequency vibrations to a responsive indicator, comprising a circuit having an input and an output, means for applying to said input waves representing sound frequency vibrations, a responsive indicator coupled to said output and means interposed between said input and output for distorting said input waves to give them an instantaneous output amplitude proportional to an approximate power function of the type produced by the normal human ear of the instantaneous wave level at the input, said interposed means comprising an amplifier together with means for positively feeding back a portion of the output energy of said amplifier to the input thereof to accentuate said distortion, and means in the output circuit of said amplifier for causing degeneration of instantaneous variations in the output of said amplifier.

2. A system for conveying waves representing sound frequency vibrations to a responsive indicator, comprising an audio frequency amplifier circuit having an input and an output, means for applying to said input waves representing sound frequency vibrations, a responsive indicator coupied to said output, and means interposed between said input and output for distorting said input waves to give them an instantaneous output amplitude proportional to an approximate power function of the type produced by the normal human ear of the instantaneous wave level at the input, said interposed means comprising circuits for feeding back to the input of said amplifier a portion of said audio frequency output energy to control the instantaneous amplitude of said modulated waves and accentuate said distortion.

3. In an audio amplifier circuit, a symmetrically distorting amplifier, a second symmetrically distorting amplifier having an anode circuit, a network which discriminates against waves of low frequency coupling said amplifiers, a feedback network which discriminates against waves of high frequency coupled to said anode circuit and regeneratively connected to said first amplifier circuit, said networks having least discrimination against waves of a low audio frequency.

4. An audio amplifier circuit comprising a symmetrically distorting amplifier having an input and an output circuit, a second symmetrically distorting amplifier having an input and an output circuit, a network which discriminates against waves of low frequency comprising the output circuit of said first amplifier and the input circuit of said second amplifier, a regenerative connection comprising a network which discriminates against waves of high frequency connected between the output circuit of said second amplifier and the input circuit of said first amplifier, said networks having a minimum discrimination against waves of a low audio frequency.

5. An audio amplifier circuit comprising an amplifier, a network coupled to said amplifier favoring a particular desired band of audio frequencies, a regenerative feedback connection deriving regeneration in part from current in the output circuit of said amplifier, and degenerative means to cause the current in the portion of the circuit from which regenerative feedback is derived to follow to a greater degree impedance variations in the final output circuit.

6. An audio amplifier circuit comprising a symmetrically distorting amplifier, a network coupled to said amplfier which discriminates in favor of a particular band of frequencies desired,'a regenerative feedback derived from current in the output circuit of said amplifier, and a degenerative connection for causing current in the portion of said output from which feedback is derived to follow more nearly impedance variations in said output circuit as a whole.

7. An audio amplifier circuit comprising a symmetrically distorting amplifler, an output circuit therefor, a plurality of networks which discriminate in favor of waves of a low audio frequency, means for regeneratively supplying signals to said amplifier through said networks, and means for accentuating in said regenerative means current variations of said output circuit due to impedance variations caused by mechanical resonance.

8. An audio amplifier circuit comprising a symmetrically distorting amplifier having an input and an output circuit, a second symmetrically distorting amplifier having an input and an output circuit, a plurality of networks connecting said amplifiers which discriminate in favor of waves of a low audio frequency, means for regeneratively supplying energy to said amplifiers through said networks, and means for providing an inverse feedback of energy to said circuits whereby current variations in said regenerative means due to impedance variation in said output circuit caused by mechanical resonance in said output circuit are accentuated.

9. An audio amplifier circuit comprising an amplifier, an output circuit therefor, a network coupled to said amplifier favoring a particular desired band of audio frequencies, a regenerative feedback connection deriving regeneration in part from current in the output circuit of said amplifier, degenerative means to cause the current in the portion of the circuit from which regenerative feedback is derived to follow to a greater degree current variations in the final output circuit, and means for preventing anode supply potential variations in said output circuit.

10. An audio amplifier circuit comprising a symmetrically distorting amplifier, a second symmetrically distorting amplifier, a network connecting said amplifiers which discriminates against waves of low audio frequency, an output circuit for said second amplifier, a network connected to the output circuit of said second amplifier for regeneratively supplying energy therefrom to said first amplifier, said network discriminating in favor of waves of low audio frequency, and means for accentuating in aid regenerative network current variations in said output circuit due to speaker resonance.

11. In an audio amplifier circuit, a symmetrically distorting amplifier, a second symmetrically distorting amplifier having an anode circuit, a network which discriminates against waves of low frequency coupling said amplifiers, a feedback network which discriminates against waves of high frequency coupled to said anode circuit and regeneratively connected to said first amplifier circuit, said networks having least discrimination against waves of a low audio frequency, whereby said audio amplifier circuit will simultaneously amplify a high frequency wave only slightly more in proportion if of a low signal level than if of a high signal level, will enormously amplify a low frequency wave of low signal level, will amplify a low frequency wave of high signal level no more than a high frequency wave, and low frequency waves will be sufficiently distorted to produce odd harmonics.

12. In an audio amplifier circuit, a symmetrically distorting amplifier, a network which discriminates against waves of low frequency coupled to said amplifier, a feedback network which discriminates against waves of high frequency coupled to said amplifier, said networks having least discrimination against waves of a low audio frequency, whereby said audio amplifier circuit will simultaneously amplify a high frequency wave only slightly more in proportion if of a low signal level than if of a high signal level, will enormously amplify a low frequency wave of loiv signal level, will amplify a low frequency wave of high signal level no more than a high frequency wave, and low frequency waves will be sufficiently distorted to produce odd harmonics.

FRANCIS H. SHEPARD, JR.

Images