US3681705A

US3681705A - Circuit and method for enhancement of signal-noise ratio

Info

Publication number: US3681705A
Application number: US889179A
Authority: US
Inventors: Lewis C Spence
Original assignee: Individual
Current assignee: Individual
Priority date: 1969-12-30
Filing date: 1969-12-30
Publication date: 1972-08-01
Anticipated expiration: 1989-08-01
Also published as: DE2064369A1; CA964336A

Abstract

Non-linear elements utilized as amplifying devices normally are constructed to have an amplification factor as linear as possible. Linearity is not achieved and a predictable distortion is generated normally mixing with the noise and signal output of the device. This invention opposes an amplified signal with a like amplified inversion thereof to cancel the linear portions of amplification and sums the non-linear portions in a common output at doubled frequency which contains the modulation information in a form enhanced relative to the noise due to the exponential character of the amplification distortion component. Noise generated or received, when of comparable magnitude to signal, is mixed with the signal input to augment the amplitude of the nonlinear amplification components notably at the crests in a doubled frequency signal from a pair of like amplifiers energized at a common point and excited in opposite phase. Noise, when generally continuous, thus enhances both the signal output and the signal-noise ratio in the output.

Description

United States Patent [151 3,681,705 Spence [4 1 Aug. 1, 1972 [s41 CIRCUIT AND METHOD FOR 2,835,806 5/1958 Verster ..328/20 ENHANCEMENT OF SIGNAL-NOISE 2,777,066 l/ 1957 Brockman ..328/20 RATIO [72] Inventor: Lewis C. Spence, 631 North 0" St., Lake Worth, Fla. 36460 [22] Filed: Dec. 30, 1969 [21] Appl. No.: 889,179

Related US. Application Data [63] Continuation-in-part of Ser. No. 815,255, April [52] US. Cl. ..330/149, 307/220, 328/20, 328/26 [51] Int. Cl ..H03f 1/26 [58] Field of Search ..330/149; 328/20, 26, 162, 163; 307/220; 325/474-476 [56] References Cited UNITED STATES PATENTS 3,091,735 5/1963 Moore ..325/476 X 2,287,334 6/1942 White ..328/162 X 3,176,239 3/1965 Browder ..330/149 3,325,739 6/1967 Stephenson ..307/220 X 3,335,290 8/1967 Fischman et al. ..328/20 X 3,261,991 7/1966 Lash ..328/162 X 3,398,297 8/1968 Huen ..328/20 X Primary Examiner-Roy Lake Assistant Examiner-James B. Mullins Attorney-Beveridge & De Grandi [5 7] ABSTRACT Non-linear elements utilized as amplifying devices normally are constructed to have an amplification factor as linear as possible. Linearity is not achieved and a predictable distortion is generated normally mixing with the noise and signal output of the device. This invention opposes an amplified signal with a like amplified inversion thereof to cancel the linear portions of amplification and sums the non-linear portions in a common output at doubled frequency which contains the modulation information in a form enhanced relative to the noise due to the exponential character of the amplification distortion component. Noise generated or received, when of comparable magnitude to signal, is mixed with the signal input to augment the amplitude of the non-linear amplification components notably at the crests in a doubled frequency signal from a pair of like amplifiers energized at a common point and excited in opposite phase. Noise, when generally continuous, thus enhances both the signal output and the signal-noise ratio in the output.

PATENTED 1 I973 saw 1 [1F 3' FIG. I

R &

WW AM w ma w u i A W 4 ATTORNEYS PATENTEnAus 1 ran SHEET 2 OF 3 FIG. 7

. INVENTOR uiwls 0 5 m ATTORNEYS PATENTEDAUB' H912 I sum 3 UF 3 B mm RV Z Mm Km 6 RM 6 I 5 MT 48 F PU Dl WW M 0 5 0 48 D 2 7 ww m E NW T 3 A R m 0 0 0 0 9 0 7 6 w W m w l 0 ma Sa o mEZ Em 0 .I n P. MD 4 0 Fr H .L. 7 v l c B R I 2 F R M l NUT mm WT CN K R 7 MW m 1 w X m T m L w w w w w 355 0 2 400 -90 -60 INVENTOR LEWIS c. SPENCE BY b 9 f FM RECEIVER INPUT SIGNAL LEVEL F l6. l0

ATTORNEYS CIRCUIT AND METHOD FOR ENHANCEMENT OF SIGNAL-NOISE RATIO This invention is a continuation-in-part of an application, Ser. No. 815,255 filed Apr. 3, 1969, relating to a Linear Expansion Amplifier.

The invention provides enhancement of a signal/noise ratio by increasing the signal energy portion of an energy-bandwidth relationship in a receiver. More particularly, it provides for expansion of the amplified product in an amplifier in accordance with the non-linear transfer characteristic of a pair of like amplifier elements excited oppositely with common output and in accordance with the broadband noise present, the noise being also enhanced, but largely filtered from the output. Transistor, diode and tube triode elements exhibit output vs. input curves which are generally exponential, usually approximately parabolic. This invention applies the outputs from oppositely phased inputs in opposition to each other, thus cancelling all like (linear) components to leave a product which is an exponential function of the input voltage variation curve, hence a means of multiplying the ratio of signal or signal plus noise to noise. It is used characteristically as a stage in the intermediate frequency portion of a receiver, but is applicable to signal processing generally, and may be used in cascaded stages of signal/noise enhancement either for a signal of the same order of magnitude as the noise or of a higher order of magnitude.

Efforts of recent years to improve the capacity of a communication system have been directed to the improvement of the noise figure of amplifiers and to means for improving the signal/noise ratio of various devices such as complete amplifiers and transistor elements. A careful study has been directed to the utilization of band widths more efficiently, to secure greater information content in a signal of specified bandwidth, since these are the factors which ultimately determine the rate and/or range of communication. Fundamental to these investigations is the consideration of noise in an input signal, noise generated in the device, noise attenuation by band limiting, and multiplexing of signals, as in digital degrees of modulation, to enhance the signal/noise ratio when factors such as bits/sec. and bandwidths are considered.

A basic treatment of the history of the development of information theory, entitled Symbols, Signals, and Noise by J. R. Pierce, Harper Bros., New York, 1961, includes works of Kelvin, Edison, Fourier, Nyquist, Hartley, Wiener and Shannon. A further theoretical treatment of noise is provided in an article at pages 539-541 in Encyclopedia of Electronics published by Reinhold, 1962. The latter publication discusses noise figures and noise temperature of a device such as an amplifier with and without efforts at noise attenuation. Such noise attenuation is there stated to be possible only with respect to excess noise above the thermal noise. In the present invention there is enhancement of the signal/noise ratio by providing a signal enhancement resulting from thermal noise, in which the thermal noise can be removed, to leave the desired enhancement. Noise ordinarily encountered in a communications receiver may contain large isolated peaks or spikes of voltage as well as a more continuous background of noise partly generated, for example, in a preceding stage. Particularly in frequency modulation detection it is possible to attenuate sporadic noise spikes by the usual limiting process. The signal information wave itself, whether of frequency modulation or other type, will contain a more or less steadily developed noise voltage superimposed upon the desired signal voltage. This invention is effective in subduing of this generally continuous noise background sometimes referred to as grass in a representation of signal plus noise voltage curves.

Theoretical treatments of signal and noise may difier somewhat in characterization of the noise as to continuous, sporadic, thermal or inherent. Regardless of theoretical treatment, it is recognized as a practical factor that background noise tends to have a relatively continuous nature and a magnitude such as to limit the range of effectiveness of information communication to a threshold generally some 3-10 decibels above the background noise. There is the further problem in any practical receiver device of avoiding or overcoming noise generated in the receiver components, which noise may be amplified along with the signal even though band limiting excludes incoming noise not passed by the input filtering system. Improvement in communication effectiveness has been limited by background noise since the signal to be processed must exceed the noise figure by a factor not heretofore reducible.

While it has been recognized that a satisfactory ratio of signal to noise is achieved by increasing the power employed in the transmission of an intelligence signal modulated upon a carrier, it has also been recognized that practical limits on the amount of power employable have been reached in numerous classes of service, oftentimes, by spill-over from one channel to another, and often because of the non-availability of higher power as in communications from deep space.

It is accordingly an object of this invention to improve the ability of a receiver to accept and process signals which are smaller in respect to background noise than heretofore utilized.

Another object of the invention is to make use of the background noise itself for enhancement of a signal representing the information content of the transmitted signal herein to achieve an improved signal/noise ratio.

Another object of the invention is to provide a means of utilization of the non-linearity of amplification of a non-linear device in a manner to improve rather than degrade the signal therein processed.

A final object of the invention is to provide an improved means of signal frequency doubling in which the doubled frequency signal contains all the information of a suppressed input signal.

These and other objects will be appreciated as the invention is described in connection with the following figures in which:

FIG. 1 is a circuit diagram of the basic circuit according to this invention;

FIG. 2 is a circuit diagram of similar apparatus with tuned input and output circuits;

FIG. 3 is a circuit diagram of similar apparatus with alternative biasing and input arrangements;

FIG. 4 is a circuit diagram of further alternative input and biasing arrangements for apparatus according to FIG. 1;

FIG. 5 is a wave form diagram of the equal and opposite inputs to balanced amplifier elements times a linear amplification factor;

FIG. 6 is a wave form diagram showing actual amplification in balanced transistors and the summed nonlinear resultant in the output;

FIG. 7 is a wave form diagram of the output of FIG. 6 with non-linear amplification products from noise spikes superimposed;

FIG. 8 is a typical characteristic curve for a nonlinear device such as a transistor; and

FIG. 9 is a diagrammatic showing of the improvement in decibels in the signal/noise ratio according to this invention;

FIG. 10 is a further graph showing the improvement achieved by this invention; and

FIG. 11 is a schematic circuit diagram using diode elements rather than amplifier elements according to this invention. 6

Previously mentioned advantages of this invention relate to the improvements indicated in FIGS. 9 and 10 which are achieved by a circuit basically having the components shown in FIGS. 1 and 11. Doubling a signal frequency is illustrated in FIG. 6 wherein a nonlinear amplification component of an amplified signal containing information to be communicated is contained in a residual signal representative of the nonlinearity from which linear amplification products are removed.

According to this invention the information contained in the input signal is assumed to be impressed upon a carrier wave, of some Kind, being typically an intermediate frequency as developed in a radio receiver after heterodyning and before detection or discrimination, but it will be understood that either an AM or FM receiver may be benefitted by this invention by the insertion of a stage such as in FIG. 1 in the intermediate frequency portion of the receiver circuit. It may be otherwise applied, whether or not one is concerned with the radio signal or other land line signal containing a data train in which signaling speed and bandwidth limitations cause the signal/noise ratio to be the limiting factor in the transmission.

Regardless of the application, a device according to FIG. 1 consists of a transformer input where a transformer generally shown at 10 has an input winding 11 with an assumed input at a frequency of F and a secondary winding comprising

equal halves

12 and 13 in series and having a center point 14 brought out to ground or to other reference voltage suitable for biasing transistors 0 and Q of which the emitters are connected together at a point 15 grounded at point 16 or equivalently connected to a reference point at other than ground. Transistors Q and Q are connected to opposite ends of windings l2 and 13 such that the voltage on transformer 10 is applied equally and oppositely at all times to the bases of Q and Q The collectors of Q and Q are connected together directly so as to completely cancel all input signals which are alike in the two transistors and all output amplification common in the two transistors. Differing components are not cancelled, but are summed.

To achieve this result, it is ordinarily preferable to provide a balancing means such as potentiometer R having a movable arm for making connection at a selected point 15 to ground at 16. This potentiometer is connected directly to the two emitters of the transistors and serves to equalize the outputs at the collectors at below threshold signals by taking account of small differences in the inputs and/or parameters of transistors of like type and manufacture. Point 15 might be selected in various ways to indicate the point at which an infinitesimally small input signal would have equal outputs at the collectors to effect complete cancellation thereof.

In FIG. 1 non-linear devices 0; and Q are illustratively transistors of identical characteristics, being shown as of P-N-P type. It will be understood that transistors of the N-P-N type may be used in an equivalent circuit with suitable changes in biasing and voltage supply. It will also be understood that Q and Q may also be other than transistors but are assumed to be devices in which amplification is not linear with changing magnitudes of input signal at the bases of Q and Q FIG. 1. Typical characteristic curves showing the non-linearity of such amplifying devices may be drawn in a number of ways of which one example will be later described in connection with FIG. 8. Triode tubes may substitute for Q and Q with suitable well known changes in connection, it being understood that the characteristic curves for such tubes also exhibit a non-linearity similar to that illustrated in FIG. 8 when large signals are not employed.

As herein described elements Q and Q are referred to as transistors or other amplifying elements, but it may be seen that the amplification factor may be less than unity, or absent, so long as it is understood that the non-linear curve of output vs. input is of exponential or logrithmic form, and the term amplifier element includes a two terminal device such as the diodes D and D of FIG. 11.

When emitters of Q and Q are connected together and grounded at point 15 and the transformer center point 14 is suitably biased by a pair of biasing resistors R and R connected in series between a voltage supply and ground. Point 14 and may be connected to the junction of R and R may be bypassed for signal by a capacitor 17 to thereby establish a signal voltage about center point 14 such that signal to the amplifiers Q and O is at all times equal and opposite with respect to a ground or arbitrary bias level voltage. In such an arrangement linear amplification at Q and Q would result in output voltages which are always in equal magnitude and opposite sign to produce an output always at zero. An output is nevertheless obtained, as will be described.

Common load 18 connects from a voltage supply 19 to a common connection between the collectors of Q and Q It will be seen that any differences in output current drawn through the collector-emitter of Q and the collector-emitter of Q between their common connection and the ground point at 15 will result in a net voltage change thereby developed across load impedance 18. This becomes a net output of the circuit taken between

terminals

20 and 21, one of which may be grounded as at 16 and the other of which may be connected to the common collector point by way of coupling capacitor 22. Further explanation of the nature and effect of an output as at 20 will be described in detail in connection with FIGS. 5 8. A sine wave input signal gives an output approximating a sign wave of double the frequency of the input to transformer 10, and as indicated at F The essential feature thus far described provides that equal and opposite phases of an input signal feed a pair of like amplifying devices which operate in the linear region of their characteristic curves to produce outputs which are alike and opposite within a first order of magnitude, to thereby present no output across the load impedance. Non-linearity of arm plification results in a second order effect from Q and Q which add in impedance 18 to provide an output at 20 which represents all frequency and amplitude components of the input signal but is of doubled frequency.

FIG. 2 illustrates an output circuit coupling to

terminals

20 and 21 utilizing an inductive coupling 23 from a tuned output circuit 24 which substitutes for a simple impedance or resistor element 18 of FIG. 1. An

output comprising coupling

23 and 24 may thus be sharply tuned to the frequency F which is also effective in removing amplified noise voltage other than components approximating F Similarly, the input circuit transformer 10 may have a tuning capacitor 25 connected across outer ends of coils l2 and 13 to provide a more precisely tuned input at frequency F FIG. 3 illustrates an alternative arrangement of FIG. 1 in which balancing is achieved somewhat differently. The input circuit is also illustrated differently, comprising means for applying F and an inversion thereof directly to the bases of Q and Q without use of a center-tapped transformer as in FIGS. 1 and 2. In this modification an inverter device 26 produces a mirror image of F being thereby also at frequency F but at a phase relation displaced either positively or negatively by 180. The input signal in this case is biased as by

resistors

27 and 28 junctioned at the base of Q and isolated from the signal source by way of capacitor 29. The base of Q is biased as in FIG. 2 except that one biasing resistor illustrated at FIG. 3 is made adjustable for the proper balancing of inputs to Q and Q The inverted image of input signal F 1 is applied by capacitor 31 at the base of Q The emitters of Q and Q may be separately connected to ground by resistors 32 and 32', one of which may be made adjustable for further balancing of combined signal output at the connected collectors of Q, and Q2 for near zero signal input.

FIG. 4 illustrates a further modification of a basic circuit in which a phase splitter device 34 provides two like signals of equal amplitude and opposite phase being passed to the bases of Q and Q by way of

attenuators

35 and 36 for balancing purposes. The FIG. 4 circuit is otherwise like that of FIG. 3 in providing biasing for transistor bases but differs in the connection of the emitters in that no resistor elements are required, the emitters being connected directly together and to an output or references such as ground. This circuit differs in illustration but not in mode of operation in that both the input and the output of Q and Q are zero. Since it is assumed for FIG. 5 that amplification is completely linear at all times, it will be evident that a suitable scale factor changes Qm and Q', to Q0: and Q'm. 6

FIG. 6 illustrates output curves at Q and Q' as of nearly sine wave form corresponding to a sine wave input as in FIG. 5 and drawn to the same scale as in FIG. 5 except corresponding to an amplified output. In an actual amplifier device employing a non-linear transfer characteristic needed to effect an amplification it will be apparent that any non-linearity in this amplification will result in a deviation of the output waves from the form of waves 40 to 411. This ideal form is' shown at 44 and 45, respectively, and actual output is shown at 46 and 47. In an amplifier device in which the characteristic curve is concave upwardly (an exponential relationship) it will be apparent that an increased signal from a static or rest point results in a greater than proportional increase in output whereas a decreased signal results in a smaller than proportional decrease in the potential at the output point. If we make a first assumption that the increase in output potential relative to the reference point above that which would represent linear amplification at that point is equal, to a first order of approximation, to the deficiency relative to a linear decrease below the static point for the output from Q, we will have curves generally as at 46 and 47, both being displaced at the maximum and minimum excursions upwardly from curves 44 and 45.

During the second half of the input sine wave the output from Q exceeds the assumed linear output by the same error as the output from Q falls short of the negative value assumed for linear amplification. The signal input at

points

42 and 43 is zero for both transistors and is accordingly zero in the output. If the deviation from linearity in each case is considered to be an error signal taken with respect to the theoretically perfect amplification of curves 44 and 45 it will be observed that the errors in output from Q and Q are additive across the load impedance of the unit as previously described. Thus an output equal to Q0: plus Q',,,,, is additive at the maxima and minima of curves 44 and 45 and zero at the axis-

crossing points

42 and 43. This is illustrated in FIG. 6 as successive maxima 48 and 49 in the resultant curves of Q plus Q' An upwardly concave, or exponential, characteristic curve for a pairof like amplifying or other devices within the so-called linear region of the characteristic curve results in additive error voltages in a circuit according to FIGS. 1 4 and 1 1. This curve is illustrated at 50 and closely approximates a sign wave of frequency F being the sum of the error components of amplification in two oppositely excited like-connected identical devices.

It will be observed that curve 50 is not an attenuated or amplified representation of the information-bearing signal at the input and in fact is of :a different frequency. Nevertheless, this output signal, which represents error from linear in amplification is always proportional to the deviation of input signal from a reference point which is the mean voltage of the input signal. This remains true whether the intelligence signal contained in the input wave is impressed thereon as amplitude, frequency, or phase modulation. Output is thus a function of the non-linearity of transfer and will be always present and additive in the output from the two devices having like characteristic curves of continuously increasing slope when output is plotted as an ordinate against input as an abscissa. Such a curve is illustrated in FIG. 8.

In constructing the curves at FIG. 6 it is assumed that the amplification of a signal containing information to be communicated is an AC input voltage of sinusoidal form and that the deviation from linearity at the crest of the output wave is approximately 2 percent in error, being a positive increment in the output voltage while the deviation from linearity at a wave trough is also about 2 percent less than would be the voltage deviation for linear amplification, the two deviations from linear adding to 4 percent in the output, depending on the magnitude of the input signal and the region of the characteristic curve into which the amplifying devices are biased for their operating points. Since the form of the input wave is not critical any form of periodic voltage reversal signal is included in the term AC input.

Characteristic curves which are continuously concave upwardly as in FIG. 8 are normally biased far from the 0 of the curve for which the non-linearity would be greatest. In any attempts at linear amplification it is of course necessary to avoid the region of saturation corresponding to high currents, and also ordinarily involving the possibility of burn-out of the devices. The regions of interest in the characteristic curve are accordingly selected within the more linear region, but such regions are found to exhibit the continuously increasing slope of the characteristic curve such that the effective error is of unitary sign deviating upwardly from an operating line corresponding to the amplification factor at the operating point. A common emitter configuration is illustrated but it will be apparent that other amplifying configurations will produce the summed error output for a characteristic curve of continuously increasing slope. Many such devices exhibit a non-linearity which is approximately a squared function corresponding to a parabolic curve within the region of interest. This factor is utilized in description of FIG. 7 hereinafter. Some other exponential function than the squared function would, of course, be effective in this configuration.

In modern high gain amplifiers the main portion of output is a linear multiple of input, corresponding to a linear equation y =f(x) B X C, where C is a constant and B is an amplifying factor. The non-linear portion may be considered as of the form of A x A being much less than B and X the principal non-linear component. In this relationship there may be other terms of lesser importance or X may be replaced in some devices with X where Z is an exponent usually approximating 2. The effect of cancelling the linear product is to remove the BX term from Y AX BX C to give AX C, C being removed in a conventional manner, so that the output is a nearly pure exponential function of the input. Since this is true for amplifiers encountered in practice we have an effective means of enhancing all larger signals relative to smaller signals present in the input, approaching, for example, 60 db relative improvement when a signal substantially larger than noise is present. The gain is not limited to signals already larger than noise, but may be used to initially render the information larger than the noise background for typical noise types, as later described.

It may be observed that curve 50 contains all of the frequency and amplitude information present in the signal F Errors assumed to be each 2 percent are added to provide an output at 4 percent of the amplitude of the sine wave perfectly amplified are shown somewhat exaggerated for clarity. This output signal may be again amplified as may be desired to reach the desired signal level.

Referring to FIG. 7 a resultant signal 50 is shown independently of the input signals and their opposed am plified voltage levels for clarity of description of the combined noise and signal. The effect of noise added to the information signal is illustrated as isolated spikes wherein the noise signal is assumed to he sets of positive and negative voltage excursions which in practice may approximate a spurious continuous signal of a much higher frequency. Successive zero output voltages are shown at 42 and 43 and successive error signal peaks are shown at 48 and 49 corresponding to the result of passing a signal wave into the circuit and taking output at 20. It will be apparent that any noise voltage input will have an effect superimposed upon curve 50. If it is assumed that a positive input voltage is supplied for a brief interval at output wave point 48 and that this input voltage is equal in magnitude to the voltage of a signal crest in curve 40, this will result in doubling the signal to be amplified as applied at the base of a transistor Q and that the effective signal excursion will rise considerably more than in proportion to the base voltage rise.

If the characteristic curve exhibits a squared relation, as is found to be approximately true even for commercially available high performance transistors, the error will be four times as large, rather than only twice as large, corresponding to the doubled input voltage. In connection with FIG. 7 we are considering the cumulative result of error in Q and Q in order to correspond I with the cumulative result shown in curve 50. If we now assume that each positive spike added to the input wave has a corresponding-like negative spike, there would be added to the error voltage for this negative spike the result of decreasing the input voltage from a crest to zero. This of course results in cancellation of the error signal since no input voltage is present at that instant. This is shown by the spike 51' extending downwardly from crest 48, this spike terminating at the zero level for curve 50. Similarly, a spike occurring at crest 49 results in an error signal 52 for a positive spike and an error signal 52 for a negative spike, both of these error signals corresponding to like noise spikes of the same magnitude as the amplitude of the input intelligence-modulated carrier. If we now consider like noise spikes at point 42 on the curve 50, we may represent a positive noise spike at 53 and a negative noise spike at 53' which are now of like result in the output, and at point 43 there is shown a similar positive noise spike 54 and a corresponding negative noise spike 54'.

If noise spikes closely follow each other and are of alternatively positive and negative sign, the results of these noise spikes will be integratively additive to the error signal shown at 50. It has been observed that internally generated noise in a system, and much of the externally applied noise, is generally continuous enough to be treated qualitatively as though it consisted on high frequency added input signals to the device.

A high frequency output superimposed upon curve 50 may be filtered or integrated as desired to emphasize the frequency F as at 50. Integration of signals 51-51', 52-52, 53-53, 54-54, and other spikes as 55-55 occurring at intermediate positions along curve 50 results in a curve substituting for curve 50 wherein the peaks are enhanced in magnitude and the troughs at 42 and 43 are unaffected. Since 51-51 illustrates the combined sine wave error signal 50 plus the noise spikes a mean value at 56 represents the instantaneous output of a transistor pair as described. Points 57, 58 and 59 similarly represent curve 50 enhanced by the noise spike average when equal and of opposite sign so that a smooth curve may be drawn to represent actual output in the presence of noise. Such a curve is shown at 60 and 61 having crests at the mean points 56 and 57 and troughs at the

mean points

42 and 43. While only a few noise spikes are shown, a more or less continuous noise background of this assumed relative magnitude would have points 62 corresponding to spikes 63 closely spaced together to give a continuous curve.

Whether or not the noise input or the internally generated noise is of white or continuous nature it is seen that the effect is to increase the magnitude of the resultant signal 50 to some new value such as 60. This signal now represents the information contained in the input signal enhanced by the presence of the noise in that a larger signal representing the input information is available.

In general the curve 60 is approximately sinusidal and is made more nearly so by the presence of filtering or tuning of the output. The noise itself may be eliminated, insofar as it is filterable or tuned out from the doubled frequency wave, while the signal representing the information average has been doubled by the presence of the noise signal of like average amplitude. If the output from a transistor pair as described is taken without filtering to eliminate the noise, there still remains an enhancement of signal/noise ratio inasmuch as signal at the doubled frequency range is enhanced while the noise is not so enhanced.

FIG. 8 illustrates a typical transfer characteristic 64 for a three element device such as a triode or transistor. It is an exponentially varying relationship of output current to input

voltage

65 and 66, being at the extremes of operation, 67 being a point below which input voltage is not normally taken because of the small response in output relative to input, the region 69-70 being in the so-called linear amplification region, still subject to the exponential relationship. A very small region such as 68 corresponds to very small signal deviations, but remains subject to the upward concavity of the curve. However, it is a signal magnitude which might be adopted for first balancing the elements Q and Q the operating static point being therein.

FIG. 9 represents at 72 relative input and output relations for a receiver including the present invention incorporated in the IF. strip, assumed to be operating at 455 KHz at a bandwidth of KHz, for an input range of 37.5DB and an output range of 73 DB between

knees

73 and 74.

FIG. 10 shows'a typical graph of noise quieting vs. input signal level according to current commercial practice.

Quieting in DB being represented as a function of signal in DBM, a portion of the graph at 78 represents 0-20 DB of quieting for inputs varying from -I lSDBM to l05DBM. A break then occurs in the graph at 77 and a smaller rate of quieting in portion 79 to provide 50DB of quieting at -75DBM. With the signal/noise enhancement of this invention the same receiver achieves results as at 75 and 76, where SODB of quieting is available at -l 10DBM in contrast to IODB for the commercial receiver. In the portion 75 of this graph it is noted that the same 0-20 DB of quieting mentioned above is available at input levels approximately 20DB lower than for the commercial set. Point 76 represents a slight change in rate of quieting increase with increasing input, being about. 40DB at the signal level where conventional quieting first commences.

FIG. 11 is similar to FIG. 1 in result in respect to enhancement of signal/noise ratio. Diodes D and D substitute for elements 0 and Q biasing being no longer needed since D and D are two-terminal devices, balanced by R and R as in FIG. 1, and double frequency enhanced S/N output being taken at 20 as a voltage across common load 18. FIG. 11 thus provides a diode non-linear enhancer which similarly cancels the first order term (BX) wherein B is less than unity and wherein the AX term remains as the output. This circuit would normally be preceeded or followed by an amplification stage to make up the loss of effective information varying signal cancelled by the two diodes connected back-to-back.

It will be apparent that the invention may be practiced in a number of equivalent arrangements and Applicant seeks to be limited only in accordance with limitations expressed in the claims.

What is claimed is:

l. A device for enhancing the signal/noise ratio of an information-modulated signal including noise components, comprising a pair of similar elements having similar non-linear transfer characteristic curves and having an input and output connection, means applying to said elements AC input signals in equal amplitude but opposite phase, at the respective input connections, means interconnecting at a point said output connections for energization in common from a power supply, means exciting said elements at greater than a predetermined signal increment but within the region of Class A amplification, and means balancing outputs from said elements for oppositely phased inputs less than said predetemtined signal increment, wherein combined output represents inequality of amplification by said elements of signals of opposite sign from a reference level.

2. A device according to claim 1 said elements being transistors having collectors connected at said point and bases connected to receive said input in opposite phase.

3. A device according to claim 2 having transistors in common emitter configuration, said balancing means comprising relatively variable emitter impedance.

4. A device according to claim 2 said means applying input signal being a transformer having a secondary connected at either end to a said element and a centerpoint at the common bias voltage level for the transistors.

5. A device comprising a pair of non-linear elements connected in parallel for opposed excitation and adjusted for cancellation of both signal and noise in an information-modulated first wave while generating a second wave of doubled frequency containing information modulated as to amplitude and/or frequency, comprising input means for supplying said first wave at an amplitude for output in the linear transfer region in a pair of equal and oppositely phased portions, said paired like elements being amplifier means having input connections to receive said portions, respectively, means biasing said amplifier means for equal and opposite outputs for said portions of input when less than a predetermined minimum, common load element means connected to a voltage source and to output connections from said amplifying means for cancellation of said outputs therein due to like but opposite values of amplification and for summing therein deviations of amplification from said like values to provide said second wave.

6. A circuit according to claim 5, said biasing means comprising at least one variable impedance controlling relative signal levels in said amplifier means.

7. A circuit according to claim including an output circuit tuned to double the frequency of the modulated wave whereby amplified noise of differing period is suppressed.

8. In a periodic signal amplifier device. employing paired amplifying elements excited, energized and biased to operate in the linear region of a common characteristic curve exhibiting continuously varying slope, means applying an input signal of a first frequency for equal and opposite amplification in said elements at one input amplitude, common load means for energizing said elements and for mixing outputs from said elements to produce absence of signal at said first frequency for said input amplitude, said common load means being connected for response to inequalities of amplification for periodic excursions of input voltage greater than said input amplitude within said linear region, said response being of periodicity equal to the number of such excursions of input voltage regardless of sign.

9. In a device according to claim 8 said elements being transistors having bases connected to receive input of opposite sign at said first frequency, having emitters at substantially a fixed potential relative to signal, and having collectors at a common potential varied as the sum of the amplification deviations of one sign according to characteristic curve slope for transistor operation at said input amplitude.

10. A periodic signal processing device comprising a pair of like non-linear current transfer elements operated in the region of approximately linear transfer characteristic connected in a common element output configuration, common load means connected from a source of DC. to a common connection to said elements for applying static voltage to equally energize said elements, means connected to said load means for taking an output voltage corresponding to the dif-

Claims

1. A device for enhancing the signal/noise ratio of an information-modulated signal including noise components, comprising a pair of similar elements having similar non-linear transfer characteristic curves and having an input and output connection, means applying to said elements AC input signals in equal amplitude but opposite phase, at the respective input connections, means interconnecting at a point said output connections for energization in common from a power supply, means exciting said elements at greater than a predetermined signal increment but within the region of ClaSs A amplification, and means balancing outputs from said elements for oppositely phased inputs less than said predetermined signal increment, wherein combined output represents inequality of amplification by said elements of signals of opposite sign from a reference level.

7. A circuit according to claim 5 including an output circuit tuned to double the frequency of the modulated wave whereby amplified noise of differing period is suppressed.

8. In a periodic signal amplifier device employing paired amplifying elements excited, energized and biased to operate in the linear region of a common characteristic curve exhibiting continuously varying slope, means applying an input signal of a first frequency for equal and opposite amplification in said elements at one input amplitude, common load means for energizing said elements and for mixing outputs from said elements to produce absence of signal at said first frequency for said input amplitude, said common load means being connected for response to inequalities of amplification for periodic excursions of input voltage greater than said input amplitude within said linear region, said response being of periodicity equal to the number of such excursions of input voltage regardless of sign.

10. A periodic signal processing device comprising a pair of like non-linear current transfer elements operated in the region of approximately linear transfer characteristic connected in a common element output configuration, common load means connected from a source of D.C. to a common connection to said elements for applying static voltage to equally energize said elements, means connected to said load means for taking an output voltage corresponding to the difference in current in said elements, said elements being diodes connected back-to-back to said load and said means applying an input signal including a center tapped transformer secondary connected at respective ends to respective terminals of said diodes opposite said common connection.