Background of the Invention
1. Field of the Invention
This invention relates to methods of increasing the channel capacity of signal transmission systems and more particularly to methods which effect the increase in channel capacity by decreasing the bandwidth requirements for transmitting quasi-periodic signals, thus permitting a larger number of information signals to be carried on a given bandwidth signal transmission channel.
2. Description of the Prior Art
Numerous signal bandwidth reduction methods are disclosed in the prior art, attesting to economic and other advantages afforded by use of such methods. For example, if a telephone communication channel could carry twice as many conversations by using bandwidth compression techniques, the cost per unit telephone could be halved. Additionally, outlays of cash and energy by the communication company can be reduced by requiring fewer channel installations.
Examples of prior-art approaches to bandwidth compression and expansion schemes are exemplified by the following U.S. Pat. Nos. 3,061,684 10/1962 Lancor et al, 3,349,184 10/1967 Morgan, 3,351,710 11/1967 Katagata, 3,510,549 5/1970 Williams, 3,541,266 11/1970 Klayman et al.
Although prior-art bandwidth compression methods are more or less effective in accomplishing their intended purposes, the methods are, for the most part, costly and complex. Accordingly, the presently disclosed bandwidth compression and expansion system was devised to overcome the limitations of previously disclosed systems. The present invention provides methods for dividing the frequency of quasi-periodic analog information signals by an integer factor, transmitting the divided frequency signal over a conventional communication channel not a part of the invention, receiving the signal and frequency multiplying it by the same integer factor to produce reconstruction of the original information signal.
Objects of the Invention
An object of the present invention is to provide a simple, low-cost bandwidth compression scheme to reduce the bandwidth required for transmitting information signals. Another object is to provide a simple, economical means for expanding a compressed bandwidth signal to its original bandwidth. Another object is to provide a bandwidth compression and expansion system which is modularly expandable to encompass a wider range of information signal frequencies than could a single bandwidth compressor and expander system.
Summary of the Invention
Briefly stated, the present invention comprehends a bandwidth compressor and bandwidth expander for use with analog electrical information signals. For purposes of illustration, the information signal considered may be an amplitude modulated quasi-sinusoidal signal with a slowly varying fundamental frequency. Such a signal is present in human speech. In operation, the bandwidth compressor clips the information signal to produce a square wave at the fundamental frequency, and digitally divides the square wave by an integern. The integral division ratio n is preferably a power of 2, to permit division to be accomplished by one or more cascaded flip flops.
The divided frequency square wave is used to alternately gate on and off an amplifier driven by the analog information signal. Thus only selected half cycles of the information signal, time coincident with the "on" time of the gated amplifier, are amplified. The output of the gated amplifier is coupled to the input of a band pass amplifier having a center frequency close to the frequency of the divided square wave corresponding to the predominant frequency of the information signal. The action of the bandpass filter results in an output signal approximating a sinusoidal signal whose fundamental frequency is an integer fraction 1/n of the predominant frequency of the information signal. In another embodiment of the bandwidth compressor, the gated amplifier is replaced with a gated integrator, resulting in a more nearly sinusoidal signal at the output of the bandpass filter.
After transmission, the received compressed bandwidth signal is restored to yield the original information signal by a bandwidth expander, one embodiment of which will now be described. In that embodiment, the quasi-sinusoidal compressed-frequency received signal is coupled to both inputs of a two-quadrant analog multiplier. The output signal from the multiplier contains frequency components twice that of the compressed-frequency signal, i.e., equal to the predominant frequency of the original information signal, if a single frequency divider were used to compress the original signal. Coupling the multiplied-frequency square wave to a bandpass filter tuned to the predominant information signal frequency produces at the output of the filter an approximately sinusoidal, varying-amplitude signal at the predominant information frequency.
If the original signal had been compressed by the factor 2.sup.n, n such multipliers would be cascaded to comprise the bandwidth expander.
Brief Description of the Drawings
FIG. 1 is a simplified block diagram of the bandwidth compressor.
FIG. 2 is a block diagram of an improved fidelity bandwidth compressor.
FIG. 3 is a diagram of bandwidth compressor waveforms.
FIG. 4 is a block diagram of an alternate embodiment of an improved fidelity bandwidth compressor.
FIG. 5 is a block diagram of another embodiment of the bandwidth compressor.
FIG. 6 is a block diagram of a bandwidth expander.
FIG. 7 is a block diagram of a bandwidth compressor employing a plurality of frequency compression bands.
FIG. 8 is a block diagram of a bandwidth expander employing a plurality of frequency expansion bands.
FIG. 9 is a diagram of waveforms for the alternate improved embodiment bandwidth compressor shown in FIG. 4.
FIG. 10 is a waveform diagram for the embodiment of FIG. 5.
FIG. 11 is a block diagram of a yet still further embodiment of a bandwidth compressor.
FIG. 12 is a diagram of waveforms for the further embodiment of the present invention shown in FIG. 11.
FIG. 13 is a block diagram of another embodiment of a bandwidth compressor.
Detailed Description of the Invention
Referring now to FIGS. 1 and 3, an information signal f.sub.m is coupled to the input of bandwidth filter 11. The output signal f.sub.mf from filter 11 is coupled to the input of square-wave converter 12. Square wave converter 12 typically employs a Schmitt trigger circuit to produce a bistable signal f.sub.sw1 at the fundamental frequency of signal f.sub.mf. Alternatively, the square wave converter 12 may employ other commonly known techniques such as a saturating amplifier or the like to produce a bipolar signal f.sub.sw1 in synchronization with a quasi-sinusoidal information signal varying in amplitude and/or frequency.
The signal f.sub.sw1 outputted from the square wave converter is coupled to the input of frequency divider 13. Frequency divider 13 preferably comprises a series configuration of n flip-flops, where n is at least one. Thus the output signal from frequency divider 13 is a square wave f.sub.sw2 at a frequency equal to the frequency of f.sub.sw1 divided by 2.sup.n. For purposes of illustration, n is assumed to be 1, giving a division ratio of 2 and producing the waveform f.sub.sw2 shown in FIG. 3. Buffer switch 14 receives signal f.sub.sw2 as an input and produces as an output signal f.sub.BS in synchronization with signal f.sub.sw2, as shown in FIG. 3. Signal f.sub.BS is of the proper amplitude to enable gated single-phase amplifier 15 for one discrete level of f.sub.BS, and disable the amplifier for the other discrete level. During the enabling portion of f.sub.BS, amplifier 15 linearly amplifies the positive half cycles of filtered information signal f.sub.mf which is coupled to the input of the amplifier, producing amplified unipolar signal f.sub.ma as shown in FIG. 3. No signal output is produced by amplifier 15 when either the enabling signal is at the disabling level, or the signal f.sub.mf is negative. Thus the output signal f.sub.ma from gated amplifier 15 comprises a train of positive-going analog signal pulses similar to the positive pulses in signal f.sub.mf, but occuring only during every other positive signal excursion of f.sub.mf, i.e., at half the frequency of the positive signal pulses in f.sub.mf. Signal f.sub.ma is applied to the input of bandpass filter 16 which is tuned to the fundamental frequency of f.sub.sw2. Thus for the example embodiment of FIG. 1, filter 16 is tuned to one half the predominant signal frequency of f.sub.m. In general, filter 16 is tuned to a center frequency which is equal to the quotient of the center frequency of filter 11 divided by 1/2.sup.n. For the specific example of FIG. 1, the signal f.sub.of outputted from filter 16 is an amplitude-modulated, quasi-sinusoidal sine wave at half the frequency of the predominant frequency component of information signal f.sub.m.
The square wave converter 12, square wave frequency divider 13, and buffer switch 14 are illustrated as three distinct circuits for ease of explanation, and because it would be the preferred mode of operation in most applications. It is possible that all three functions could be performed by a single frequency division circuit such as a flip flop which operated with an input of certain sinusoidal type signals, and whose output keyed the gated amplification circuit on and off directly. If an input signal was of high enough quality, it would be possible to dispense with the initial bandwidth filter. The gated amplifier was described as amplifying only positive going pulses, but amplifying only negative going pulses would work similarly.
An improved fidelity bandwidth compressor is illustrated in FIG. 2. In this embodiment, an additional gated single-phase amplifier 15' is used to amplify and invert alternate positive half cycles of filtered information signal f.sub.mf in a manner now to be described. Referring to FIGS. 2 and 3, buffered square wave drive signal f.sub.BS used to control gated amplifier 15 is inverted by inverting buffer amplifier 17 to produce control signal f.sub.BS ' for controlling gated amplifier 15'. Gated amplifier 15' is a single phase inverting amplifier responsive only to positive-going input signals coincident with the positive level of f.sub.BS, inverting the positive signals to negative signals. Thus the output signal f.sub.man from amplifier 15' is a train of negative going analog signals similar to the positive excursions of signal f.sub.ma, but occurring only during every other positive signal pulse of f.sub.mf, i.e., at half the frequency of the positive signal pulses in f.sub.m.
Signal f.sub.man is summed with signal f.sub.ma by a summing operational amplifier 18. The amplitude of the output signal FA3 from the operational amplifier is given by the equation FA3=f.sub.ma (R3/RI)+f.sub.man (R3/R2) (equation 1). As shown in FIG. 3, signal F.sub.A3 comprises a train of positive and negative-going signals similar to the positive and negative excursion of signal f.sub.mf but occurring only during positive excursions of information signal f.sub.mf. Signal F.sub.A3 at the output of operational amplifier 18 is coupled to the input of a bandpass filter 16 having a center frequency equal to one-half the fundamental frequency of the information signal f.sub.mf. The action of bandpass filter 16 on signal F.sub.A3 results in a quasi-sinusoidal waveform F.sub.OT at the output of the bandpass filter. In order to divide an input signal by N=2, it is preferable to invert alternate positive going signals to negative going signals in order to get an intermediate outgoing signal with a symmetrical shape at the divided frequency. If a divider was used with an odd division ratio, such as N=3, a negative going excursion would naturally occur in the input signal at the temporal midpoint between selected positive going excursions of the divided outgoing signal. There would be no need to invert any positive going input excursions, and every third negative excursions would be amplified with no necessary phase inversion. The use of this type of division process is therefore not intended to be limited to even division ratios, but is described as such because of the operational simplicity and preferred mode of operation.
To reconstruct an information signal which has been compressed in frequency by the methods herein described, a bandwidth expander of the type shown in FIG. 6 may be employed. As shown in FIG. 6, compressed bandwidth signal F.sub.OT is applied to the A and B inputs of two-quadrant analog multiplier 20. The signal at output terminal C of multiplier 20 is given by the equation V.sub.C =kV.sub.A V.sub.B (equation 2); where V.sub.C is the output voltage, V.sub.A is the input voltage at terminal A, V.sub.B is the input voltage at terminal B, and k is the gain constant of the multiplier. Since input terminals A and B are connected together in the multiplier configuration shown in FIG. 6, input signals are either both positive (quadrant 1) or both negative (quadrant III) and therefore multiplier 20 need only have a two-quadrant multiplication capability.
Now for a compressed bandwidth signal V.sub.OT having a quasi-sinusoidal waveform, the instantaneous amplitude is given by V.sub.OT =E.sub.P sin w+(equation 3) where wt is the angular frequency and E.sub.P is the peak amplitude which, in general, is a function of time, but is slowly varying with respect to the angular frequency w. Substituting equation 3 into equation 2 gives the expression for the output voltage Eot from the multiplier: E.sub.OT =k E.sub.P.sup.2 sin.sup.2 wt (equation 4). Substituting the trignometric identity sin.sup.2 wt=1-cos 2 wt/2 into equation 4 gives: E.sub.OT =k E.sub.P 2 (1-cos 2 wt)/2 (equation 5). E.sub.OT comprises a sinusoidal signal having a fundamental frequency 2w which is twice the fundamental frequency of the compressed signal F.sub.OT with an amplitude proportional to the instantaneous amplitude E.sub.P of F.sub.OT, and a superimposed direct current voltage level. Therefore, for a compressed bandwidth signal which are originally produced by dividing the frequency of the information signal by two, the operation of the bandwidth expander shown in FIG. 6 results in restoring the fundamental frequency of the transmitted signal to its original value. The d.c. component superimposed on the recovered signal may be removed by capacitively coupling the signal to its final destination.
It is evident from the description of the operation of analog multiplier 20 that the multiplier, configured as shown in FIG. 6, functions as a frequency doubler. The frequency doubling function may also be performed by such non-linear elements such as diodes, or alternatively numerous other frequency doubling circuits are suitable for use in place of analog multiplier 20.
In general, where the frequency compressors previously described herein employ n cascaded flip-flops to divide the predominant fundamental component by a factor of 2.sup.n, the corresponding bandwidth expander similar to the one shown in FIG. 6 will require n cascaded frequency doublers 20.
Referring now to FIG. 4, another embodiment of an improved bandwidth compressor is shown. The operation of that embodiment is analogous to the operation described for the improved bandwidth compressor of FIG. 2, up to the point where gated single phase amplifier 15 and inverting single phase amplifier 15 in FIG. 2, are replaced with gated, complementary unipolar integrators, 19 and 19' in FIG. 4. As illustrated by the waveforms in FIG. 9, the function of integrators 19 and 19' is to make the waveform which is the composite sum of signals F.sub.11 and F.sub.12 outputted by integrators 19 and 19' more closely approximate a sine wave. The action of summing amplifier 18 and filter 16 following integrators 19 and 19' is exactly the same as has been described for the bandwidth compressor shown in FIG. 2.
Another embodiment of the bandwidth compressor is shown in FIG. 5. The waveforms associated with the signals of the compressor shown in FIG. 5 are shown in FIG. 10. The operation of the circuit elements in the parallel circuit paths up to the output of buffer switch 14 and to the output of bandwidth filter 11 is exactly as has been previously described for the embodiment shown in FIG. 2.
Referring now to FIG. 10, filtered information signal f.sub.mf at the output of bandwidth filter 11 is coupled to full-wave rectifier 30 producing full-wave rectified waveform F.sub.mfr as shown in FIG. 10. Signal F.sub.mfr is coupled to the input of peak detector 31, which peak detector generates a voltage proportional to the peak value of F.sub.mfr occurring during the positive half cycle of FSW.sub.2. Peak detector 31 is reset by signal FSW.sub.1 generated by differentiating the leading edge of FSW.sub.1 with differentiator 33. The output signal from peak detector 31 is gated into sample and hold circuit 32 by sample and hold signal FSW.sub.1, generated by differentiating the trailing edge of FSW.sub.1 through use of inverter 34 and differentiator 35. Output signal F.sub.S from sample and hold circuit 32 is thus a rectangular waveform unipolar signal. Sample and hold circuit output signal F.sub.S is coupled directly to gated amplifier 15, and also through inverting amplifier 36 to gated amplifier 15'. Gated amplifier 15 is enabled by buffered signal F.sub.BS in synchronization with the positive half cycles of FSW.sub.2. Thus, sample and hold signal F.sub.S is amplified during the positive excursions of FSW.sub.2.
Buffer switch signal F.sub.BS is also connected to the input of inverter 17, resulting in a positive-going output signal F.sub.BS ' during the negative excursions of F.sub.BS. Signal F.sub.BS ' is used to enable gated amplifier 15', which is driven by inverted sample and hold signal F.sub.S ' occurring at the output of inverting amplifier 36. The outputs of gated amplifiers 15 and 15' are summed by operational amplifier 18 to produce a bipolar square wave F.sub.SA. Signal F.sub.SA is in turn coupled to bandwidth filter 16 to produce filtered bipolar signal F.sub.OT.
It is also possible to use the peak detector and hold circuits on preselected pulses of the information signal, such as every other positive or negative pulse, with or without using a full-wave rectifier. The maximum or minimum peak over a desired period may be easily retained by sample and hold circuits when the full-wave rectifier is incorporated, and is the desired embodiment for descriptive purposes.
The bandwidth compression and expansion methods herein before described each employ a single bandwidth bandpass filter at the output of the bandwidth compressor. While such compression and expansion techniques are adequate for information signals comprised mainly of a fixed fundamental frequency, sinusoidal signal modulated by lower frequency components, a more general class of information signals comprises signals with a varying fundamental frequency. To accommodate the latter more general class of information signals, the embodiment shown in FIG. 7 may be employed. In the multi-channel bandwidth compressor shown in FIG. 7, a plurality of bandwidth compressors are driven by the information signal f.sub.m. For simplicity of illustration, the rudimentary embodiment bandwidth compressor of FIG. 1 is shown as the basic compressor in FIG. 7. It is obvious, however, that any of the previously described embodiments of the bandwidth compressor could be used in the technique shown in FIG. 7. Also for ease of description, it is assumed that the square wave frequency dividers 13 in FIG. 7 are single flip-flops, resulting in a frequency division by a factor of 2. Thus, if an input bandwidth filter 11 is tuned to a frequency f.sub.1, the compressed bandwidth output filter 16 is tuned to f.sub.1/2.
Now, if the information signal f.sub.m comprises a number of predominant frequency components f.sub.1, f.sub.2 . . . f.sub.n, each modulated by frequency components lower than the corresponding predominant frequency, it is clear that a plurality of bandwidth compressors as shown in FIG. 7, each of which is tuned to f.sub.1, f.sub.2, . . . f.sub.n, respectively, may be used to compress each range of signal frequencies centered about f.sub.1, f.sub.2, . . . f.sub.n. The respective compressed bandwidth signals may then be summed in an amplifier 18 to produce a composite compressed frequency signal.
In order to tune a filter to the center of a low frequency divided signal which slowly varies in frequency, it is also possible to heterodyne the information signal with a high frequency carrier signal, and filter off the desired side band. The side band can then be divided by N, and the filter tuned to the smaller bandwidth segement with the center frequency at 1/N. For example, a signal varying between 100 Hz and 1000 Hz could be heterodyned with a 10 KHz carrier, and the single side band 10.1 to 11 KHz selected. The filter for this single side band when divided by 2 could be tuned to a center frequency of approximately 5.55 KHz.
To reconstruct an information signal which has been compressed in frequency by a multi-channel bandwidth compressor similar to that shown in FIG. 7, a bandwidth expander of the type shown in FIG. 8 may be employed. As shown in FIG. 8, the compressed bandwidth signal F.sub.ot is coupled to a plurality of paralleled bandwidth filters. For each bandwidth filter in the bandwidth compressor preceding the bandwidth expander, there is a corresponding bandwidth filter in the bandwidth expander tuned to the same center frequency, for a total of n filters. Following each bandwidth filter in the multi-channel bandwidth expander is a bandwidth expander 20 identical in operation to the basic bandwidth expander of the type shown in FIG. 6, which operation has been described above. The expanded bandwidth signal from each expander 20 is coupled to an input of summing amplifier 42. Thus the output summing amplifier 42 is a composite signal comprising the sum of n expanded bandwidth signals resembling the n bandwidth signals into which the original information signal was divided before individual compression, summing and transmission to the multi-channel bandwidth expander.
When a complicated communications signal has a narrow enough bandwidth so that the signal within that bandwidth approaches a sine wave, conventional dividing techniques can be used for signal compression. However, such conventional filtering techniques can be costly in that many new and separate filters would have to be used. Consequently, such a method, while useful, has significant practical limitations. Additionally, it is further questionable that such conventional filters would be useful without electrically "matching" the characteristics of one filter to another in order to insure faithful processing of the signal waveforms.
Referring now to FIGS. 11 and 12, an important method for achieving bandwidth compression is presented by utilizing a bandwidth filtering technique which uses a single bandwidth filter (not shown) and, thereby eliminates the need for numerous different bandwidth filters. For example, if two very simple sine wave signals, .sup.F L and .sup.F H, as shown in FIG. 12, were combined, a more complicated signal .spsp.F.sup.H L would be formed. If .sup.F H was sine wave of 2,400 Hz and .sup.F L was a sine wave of 1,200 Hz and the combined signal .spsp.F.sup.H L was input to the device shown in FIG. 11, containing both of the fundamental frequencies, 2,400 Hz and 1,200 Hz and the sum and the difference of those two fundamental frequencies, 600 Hz and 3,600 Hz. These four (4) frequencies are input to the circuit shown in FIG. 11 and initially are fed into a pair of mixers 50,51 having a corresponding pair of oscillators 52,53 having frequencies of 9,000 Hz and 10,000 Hz, respectively. The output of the mixers 50,51 would be fed into the filters 54,55, which although electrically distinct, the filters 54,55, as a practical matter, might occupy a single physical structure or housing. The outputs of the bandwidth filters, which, of course, have a relatively narrow bandwidth of 1 KHz, extending from 11,000 Hz to 12,000 Hz, would be .sup.F H+9,000 Hz and .sup.F L+10,000 Hz, respectively. The output of filter 54 would, of course, be a sine wave resembling .sup.F H, but at the higher frequency of 11,400 Hz. This is, of course, due to the well known characteristic of a conventional filter, such as an LC or so-called "tank" circuit, which when a non-sinusoidal waveform is input into it, it will output a sinusoidal waveform.
Correspondingly, the output signal of filter 55 would be a sine wave resembling the input frequency, .sup.F L+10,000 Hz, but at a frequency now of 11,200 Hz. The output signals of filters 54,55 are then fed into a pair of frequency dividers 56,57 which divide the corresponding frequencies in half, that is (.sup.F H+9,000 Hz)/2 and (.sup.F L+10,000 Hz)/2, respectively, the resulting bandwidth being, in each case, 5,500-6,000 Hz. Both the output signals from the frequency dividers 56,57 are input into a pair of mixers 58,59 having a corresponding pair of local sine wave oscillators 60,61 having output signal frequencies of 4,400 Hz and 5,000 Hz, respectively. Following this, the outputs of the mixers 58,59 are input into conventional filters, .spsp.F.sup.H 2 and .spsp.F.sup.L 2, respectively. Filters .spsp.F.sup.H 2 and .spsp.F.sup.L 2 have bandwidth of 500-1,500 Hz. Consequently, the output signal of .spsp.F.sup.H 2 will be the signal .sup.F H at the divided frequency; namely: .sup.F H/2 or 1,200 Hz. Correspondingly, the output of filter .spsp.F.sup.L 2 would be the signal resembling .sup.F L at the divided frequency which, in the present case, is .sup.F L/2 or 600 Hz.
Thereafterwards, .sup.F L/2 and .sup.F H/2 are summed together in conventional fashion by the summing element 65, and the resultant signal will be .spsp.F.sup.H L/2 or .spsp.F.sup.H L compressed by a factor of 2.
It should also be noted at this time, that this method could also be used by lowering the frequency of the individual narrow bandwidth segments of signal prior to dividing or compressing the individual segments. Following signal filtering via filters 54,55, each narrow bandwidth segment could be mixed with the desired carrier, and a lower frequency signal sideband filtered off. Thereafterwards, the lower frequency segments could then be treated individually.
In the expansion of a compressed bandwidth signal, if the compressed signal is separated into small bandwidth segments in which the signal therein approaches a sine wave, frequency multipliers such as a two quadrant multiplier or a full wave rectifier can be used on the individual bandwidth segments with minimal operational error. The frequency multiplied bandwidth segments can then be recombined or summed to form the expanded bandwidth signal.
For economy in producing narrow bandwidth segments for frequency multiplication or bandwidth expansion, the technique of using several individual, but electronically identical filters can also be used in the same manner as illustrated in compression.
Each filter could select a separate portion of the wider bandwidth signal by using a unique carrier or mixing signal prior to each filter. Each filter could then filter off a single side band which would be a unique portion of the original signal. The narrow bandwidth side band could then be multiplied.
In many cases it would be desirable to multiply this segment several times if there was an original compression factor greater than two. Between each multiplier it would be desirable to use a bandwidth filter. In the case where the signal is initially multiplied by two, a bandwidth filter tuned to a center frequency of approximately twice that of the signal before multiplication should be used. This would reconform the multiplied signal to a quasi sine wave for input to the next multiplier, and would eliminate intermodulation products which might otherwise develop.
With reference now to FIG. 13, as yet still another embodement of the present invention is shown and described. Basically, this circuit generally shown at 66, employs the theory or methodology that the separation of a complicated communications signal into small bandwidths insures that a close approximation of a sine wave can be ultimately achieved. Should the signal approach a true sine wave, conventional methods of converting the sine wave into a square wave and subsequently frequency-dividing the square wave would work well in frequency division methodology. However, if the signal is not a close approximation of a true sine wave, such as might be the situation if two elementary sine waves were combined, the signals would add and subtract in a manner which would not permit a conventional frequency divider to operate properly at all times. The reason is understandably simple when one realizes that as the amplitude of -one of the signals is varied with respect to the other signal, at frequencies of, say, 1,500 Hz and 1,000 Hz, false zero crossings might result. If such would be the case, then the conventional frequency divider seeing the two signals summed together at 2,500 Hz might appear large enough to produce false triggering of the conventional flip-flop, or initially one signal might appear larger, and then the second signal, or even the difference frequency of 500 Hz might cause false triggering.
The instant method proposed by the present invention herein is to hetrodyne the input signal with a higher frequency carrier signal than present in the input signal, filtering off a side band, and then dividing or compressing the side band signal bandwidth.
After the side band bandwidth is compressed, it can then be again returned to the frequency range by once again hetrodyning the compressed bandwidth with an appropriate carrier and filtering off a sideband corresponding to the signal resulting from directly compressing the bandwidth of the original input signal.
For example, a complicated signal of two elementary signals with a fundamental frequency of 1,000 Hz, which is varied in amplitude, could be hetrodyned with a 10 KHz carrier signal and the single upper side band filtered off. The resultant signal would appear to be a closer approximation to a sine wave, with the information signals at 1,000 Hz and 1,500 Hz causing some distortion, relating to the frequencies of 11,000 Hz and 11,500 Hz. Even if an additive frequency of 12,500 Hz occurred, the resultant signal would not cross the zero level significantly more often, and cause extreme triggering of the frequency divider, since ordinarily the number of zero crossing would be 11,000 to 11,500 Hz per second anyway.
This technique is illustrated in FIG. 13 where input signal .sup.f in is mixed by a mixer 67 with a carrier signal generated by an oscillator 68 resulting in output signal .sup.f mix. A bandwidth filter 69 then selects a single side band .sup.f ssb which corresponds to .sup.f m in FIG. 2 which is basically similar in the remaining part of the circuit shown in FIG. 13. The signal wave forms output from the bandwidth filter 69 are the same as shown in FIG. 3. Signal .sup.f ssb then is converted by a square wave converter 70 to a square wave .sup.f swl and then divided by the square wave frequency divider 71 to .sup.f sw2. Square wave divided signal .sup.f sw2 activates the single phase gated amplifier 72 during one phase, and then keys it off in the opposite phase. In turn, .sup.f sw2 keys the single phase inverting amplifier 73 on when the first amplifier 72 is off, and off when the first amplifier 72 is on. The single phase gated amplifier 72 outputs signal .sup.f ma and the single phase gated inverting amplifier 73 outputs signal .sup.f man. Signals .sup.f ma and .sup.f man are then summed by the summing element 74 to form signal FA3 which is then filtered by the bandwidth filter 75 to form signal FOT.
This technique allows using larger bandwidth segments to efficiently operate with a square wave frequency divider. As the bandwidth segment becomes larger the importance of hetrodyning the signal with a carrier signal and obtaining a higher frequency single side band becomes increasingly desirable to insure proper operation of the square wave frequency divider.