Outputs 5000Hz high frequency output is fed to point A to establish the primary frequency, and a comparatively lower frequency signal is fed to point B to establish a desired offset from the primary frequency. For ex- ample, a 50.000 kHz signal to A will cause lockup as noted above. If a 1000 Hz signal from the stan- dard is fed to point B, the summed signals will be at 14,051.000 kHz and 14,049.000 kHz. These will produce the desired 1 kHz offset in either the l.s.b. or u.s.b. mode, delivering the required 1 kHz audio output to the c.c.w. filter. However, very careful shielding is an absolute must so that the harmonic of the 1000 Hz mixing signal from the standard does not interfere with the incoming c.c.w. signal. W6NEY reports that with his system, the lockup range at 14 MHz is about 400 Hz wide, making possible lock- up points every 1 kHz or so, and possibly even closer. Further refinements will permit operation of c.c.w. stations as close together as 10 Hz in W6- NEY's estimation. The Petit CCW Filter The Petit c.c.w. Filter consists of two filter chains operated 90 degrees out of phase as shown in fig. 9. Input to the c.c.w. filter is a 1000 Hz signal from the audio output of a conventional c.w. receiver, or from the product detector of a c.c.w receiver. Filter output is a reconstructed 1000 Hz audio signal. The 4 kHz timing signal to the c.c.w. filter is derived from the master frequency standard, and subdivided within the "filter driver" to provide the driving pulses for switches U2, U3, and U5. U9 varies the phase of the driving pulses to U3 in 0.01 second steps through switch SW1 in order to permit time synchronization with the distant keyer pulse-phase. The c.c.w. filter is the heart of the system, and an understanding of its operation is essential to an understanding of c.c.w. As can be seen from fig. 9, each filter chain consists of an input mixer, and integrator, a sample/ hold stage, and a balanced modulator. The opera- tion of the filter can best be explained by analyzing the function of each of these stages. 1.) Input Balanced Mixers. The function of the input balanced mixers is to convert the a.c. voltage audio input signal to a d.c. voltage which drives the inte- grator stage. Its operation can be explained by analogy with the center-tapped transformer, s.p.d.t. switch, and ripple filter of fig. 10. If we apply a switching signal which causes the switch to change polarity at exactly the same frequency as the a.c. input signal (1 kHz), traces a.c. show what happens relative to the phase between the input and switch- ing signals. If the phase between these signals is 0 degrees, the switch will always be connected to the side of the transformer which is swinging in the positive direction, producing an output that looks like the familiar fullwave rectifier output (a). Filter- ing produces a constant d.c. voltage which is an average of the output. In the opposite case (c), where the switching signal is 180 degrees out of phase with the input signal, the switch will always connect to the side of the transformer which is swinging negative, producing a mirror image of the positive output. However, if the input and switching signals exhibit a 90 degree phase difference (b), then the switch will connect to the opposite side of the transformer at quarter-cycle peaks, causing a cancellation and zero output after filtering. Finally, if the switching signal is drifting out of phase with the input signal, the output will not be a d.c. voltage, but an a.c. voltage or "beat note". A 1 cycle or 360 degree drift is shown in (d), with the average output after filtering zero volts. In the case of the Petit c.c.w. Filter, with its 0.1 Hz integrating interval, a signal which is 10 Hz from zerobeat will go through one complete cycle every 0.1 Hz and will produce no output (e). In short, the filter produces its greatest output for input and switching signals exhibiting a zero-phase relationship, or precisely at zerobeat, and rejects signals in proportion to their distance from zerobeat. Fig. 2 shows the frequency response of the Petit 0.1 Hz filter and a comparison of it with conventional filter bandwidths. 2.) Integrators. The output from the balanced mixers are fed to the integrator and sample/hold stages, which are the heart of the c.c.w. filter and perform the time-averaging task. The integrator sums or av- erages the signal strength at its input during the in- tegrating period (pulse length); it does not respond to short term changes in the amplitude of a signal, but averages them. The integrator is similar in func- tion to the familiar "charging capacitor." During a charging period, its charge rises to the level of the input, but unlike the capacitor, which charges along the familiar exponential curve, the integrator's charge rate is linear. Second, like the capacitor, the integrator can remain charged at a constant value during the charging period, or portion of a charging period. Third, the integrator differs from the capaci- tor in that it can be "dumped" or reset to zero charge almost instantaneously. Fig. 11 shows in- tegrator behavior. Graph (a) shows a constant d.c. voltage as the input to the integrator. For each of the four inputs signals, the integrator will charge linearly and proportionally. The summed voltage present at the end of the integrating period is the only voltage of interest, and drives the sample/hold stage, as will be noted below. Graph (b) shows integrator response to a square wave input. Graph (c) shows integrator output when the input signal goes through one complete cycle during one inte- grating period. The integrator sums the input voltage and shows zero output. Graph (d) shows a multiple- cycle per integrating period condition. Two observa- tions should be made. First, in the multiple-cycle case, the amplitude of the integrator response di- minishes in proportion to the number of cycles. Second, only the average of the voitage of the un- completed cycle will appear in the integrator output. So, four basic cases describe all integrator be- havior: 1.) constant d.c. input; 2.) square wave less than complete cycle; 3.) square wave, complete cycle; 4.) multiple of complete cycle, and multiple of "complete cycle plus incomplete cycle." As can be seen, cases 2 and 4 are variations. Fig. 12 shows the c.c.w. Filter and Driver circuits. Examination of U3 will clarify integrator switching. As can be seen, U3>~v1 (E) and U3>\8.3tF) short out Ulc/Uld integrators when the "dump" or reset pulse is applied to terminals E/F. This ends one integrat- ing period and begins the next. In other words, the integrator is left to charge during the integrating period, and the reset pulse"discharges" the inte- grator at the end of the period. With this understanding of the integrator in mind, it can be put back into the signal flow through the filter to this point. When the mixer input signal (1 kHz audio) and the mixer switching signal (1 kHz) are exactly in phase or zerobeat, the highest value of d.c. output will appear at the output of the mixer and drive the integrator, as in fig. 5a. If the mixer input and switching signals are not exactly in phase, then the output to the integrator will be a number of a.c. cycles, as in fig. 10d or fig. 11d. If this a.c. drive signal is a number of complete cycles plus an incomplete cycle, integrator output will only be the sum of the voltage of that final, incomplete cycle. It will be proportionally smaller in magnitude as its frequency increases. These principles of integrator operation allow us to understand how the filter can operate at such a narrow bandwidth of 10 Hz. With a 12 w.p.m. and 0.1 second pulse length, the filter is analyzing sig- nal inputs in blocks of 0.1 second. Thus, a 10 Hz input signal to the integrator will go through one complete cycle every 0.1 second, as shown in fig. 10e, and integrator output will be zero. if, however, we apply a 5 Hz signal, it will go through one com- plete cycle every 0.2 seconds, or two integrating intervals, or one-half cycle every 0.1 second inte- grating period, and hence, the filter response will be 6dB down. To put this in a practical context, a signal at 14,050.005 kHz will be 6dB down from a signal at 14,050.000 kHz, where the desired c.c.w. signal is located. 3.) Sample/Hold Stages. The sample/hold stage is simply a switch, capacitor, and voltage follower. Its operation is quite simple. At the end of an inte- grating period, it samples the voltage present at the integrator output, quickly charges to that level, and passes this voltage on as a drive signal to the bal anced modulators. The only integrator voltage that the sample/hold is concerned with is that which is present at the end of an integrating period. The sample/hold stage switch is closed by a .01 second pulse and samples during that duration, which is one-tenth of the integrating period. The output of the sample/hold stage which drives the balanced modulator and produces the reconstructed c.w. audio note, then, is telling us what the-output of the integrator was at the end of the previous integrat- ing period. In other words, the output from the filter lags behind the input by one integrating period, or 0.1 second. The switching relationship between the integrator and sample/hold will be noted in describ- ing the "filter driver" below. 4.) Balanced Modulators. The function of the bal- anced modulator is to convert the d.c. drive voltage of the sample/hold stage to an audio tone- which reconstructs the desired c.c.w. signal. The ampli tude of its output is directly proportional to the d.c. input voltage. 5.) Filter Driver. The Filter Driver unit is driven by a 4 kHz signal from the master frequency standard, and subdivides and phases that signal for the re- quire switching signals in the filter. The derivation and operation of these signals can be seen by reference to fig. 9. U6 divides the 4 kHz input by four and produces 1 kHz outputs exhibiting a 90 degree phase differ- ence which are used to switch, first, the input mixers via U2, and secondly, the balanced modulators via U5, to provide the output to the inte- grators and the audio output from the balanced modulators respectively. A third 1 kHz output from U6 is divided by ten in U7, and outputs to U8 are taken from the "2" and "4" decoded decimal terminals of U7. The string of pulses from U7 from the "2" and from the "4" decoded decimal terminals will be separated by the duration of the missing "3" pulse, which es- tablishes the time difference between the "sample" pulse (from "2") and the "reset" pulse (from "4") which eventually reach integrator and sample/hold switches in U3. The outputs from "2" and from "4", then, each consist of a string of pulses, 10 per 0.1 second integrating period, the sample pulses oc- curring at .002-.012-.022—.092 seconds, the reset pulses at .004-.014-.0221 .094 seconds, thereby es- tablishing the "units" sequence. Next, a 100 Hz signal is fed from the "carry" terminal of U7 to U9 and divided by ten to produce one pulse of .01 second duration per 0.1 second integrating period. This pulse may be selected from any of the ten decoded decimal terminals of U9 and establishes the "tens" sequence of pulses, with one pulse occurring every 0.1 second, but select- able as to which .01 pulse position in the 0.1 second period is used. For example, with the phase-adjust switch SW1 in the "4" position, the "tens" pulse will occur at the .04 second position, and will choose the sample pulse at 0.042 and the reset pulse at 0.044 seconds coming from U7; with SW1 at position 8, the sample pulse at .082 and reset pulse at 0.084 will be selected. This selection oc- curs in U8, which consists of a pair of cascaded dual-input NAND-gates which fire only when the decade pulse from U9 and "units" pulses from U7 appear at its inputs coincidentally. Fig. 13 should make this timing operation clearer. In short, the phase-adjust switch establishes: 1.) the length of the integrating period at 0.1 second; 2.) the precise instant at which the integrating period will begin in increments of .01 second. Finally, it should be clear that sample and reset pulses last only one-tenth of the integrating interval, the sample pulse closes the sample/hold switch of U3 while that stage charges for 0.01 second, another 0.01 second elapses, and then the reset pulse, which lasts an- other 0.01 second, shorts out and discharges the integrator. The next integrating period then begins. 6.) Filter Specs. The prototype c.c.w. filter by Petit operated with a 50 kHz input signal and slightly different timing than the filter shown. It required an input of less than 50 millivolts (c.c.w. signal) for proper operation; undesired signals (non-c.c.w.) could reach 2 volts p-p before saturating the filter. The improved version discussed in this paper, de- signed and built by WA7ZVC, saturates with a 300 millivolt p-p c.c.w. signal, but non-c.c.w. signals can reach 6 volts p-p before saturating the filter. In prac- tical terms, this means that non-c.c.w. signal and QRM levels can reach monstrous proportions be- fore causing interference to the desired c.c.w. sig- nal. Construction and adjustment of the Petit c.c.w. Filter is quite within the capabilities of the e xperienced homebrewer. Future Developments Problems encountered in initial experimentation have led to theoretical planning and work on future c.c.w. systems. The problem of harmonic interefer ence from the frequency standard is being elim inated through the choice and implementation of new c.c.w. net frequencies of 3,562.500 KHz and 14,062.500 kHz, which are relatively easy to synthe size and which avoid such interference. Close at- tention is being given to receiver design, since the c.c.w. concept pushes receiver technology to its limits. Practical efforts at present include the de- sign of a receiver specifically for c.c.w. use. It will include phase-locked HFO and BFO, 9 Mhz IF with sharp filter, front-end overload protection, an inte- gral c.c.w. filter, and a product detector for con- ventional c.w. A similar transmitter design is being worked out. The need to limit operations at this point to single net frequencies within each band (80 & 20 meters) has been eliminated by the design and construc- tion of a 5-5.5 MHz frequency synthesizer by W7GHM which delivers outputs of about 1 volt rms in 100 Hz steps across the range. A 1 kHz TTL refer- ence signal derived from the c.c.w. station standard provides stabilization. With this synthesizer, a 100 kHz wide hamband will allow 1000 c.c.w. channels! Further improvements in the basic Petit Filter and receiver schemes will enhance the flexibility and practicality of c.c.w. operation. The Future of CCW and Amateur Radio C.c.w. appears to offer an obvious solution to the problem of crowded amateur bands. First, by re- quiring such a narrow bandwidth, a far greater number of c.c.w. stations can occupy each fre- quency segment. Secondly, because of the great improvement in signal-noise ratio offered by c.c.w., stations will require far less power to maintain communications. With c.c.w., a permissible reduc- tion of about 20dB in radiated power could make the QRM-less situation a "dream come true." Final- Iy, with dark clouds gathering on the horizon in the form of the upcoming WARC in 1979 and prognos- tications of the possible loss of large chunks of the amateur bands to other interests, radio amateurs may be forced to turn to c.c.w. to utilize the space that is left, as a loss of spectrum is inevitable at WARC. At present, c.c.w. is just getting off the ground; with stations in operation at W6NEY, W7GHM, and WA7ZVC, and activity preparing on the East Coast, Europe, and Japan, and hopefully, elsewhere after the appearance of this paper, but theoretical plan- ning and work on future c.c.w. systems is underway. The major thrust of this work is to increase the flex- ibility of the c.c.w. concept to where it is no less flexible than conventional c.w. The necessary tech- nology exists and only requires application. W6NEY is working on a conversion of an HW-8, a superb piece of gear, to c.c.w. and envisions an entire c.c.w. station, minus frequency standard, inside the cabinet. Next, perhaps, comes a digitally synthe- sized c.c.w. transceiver (output 5 watts or less, who will need more?) capable of operating at 100 Hz points through the amateur bands and literally working the world from a lantern battery supply and simple antenna under even poor propagation conditions. C.c.w. is the mode of the future! Conclusion In closing, l must emphasize my complete debt to W6NEY, W7GHM, WA7ZVC, and the CCW News- letter, for the information included in this paper. Time has prevented them from writing such a paper, and it is my judgment that knowledge of c.c.w. is in the best interests of amateur radio. I would like to thank W6NEY for reading and making corrections to this paper. Further information can be found in the Coherent CW Newsletter (CCWNJ, edited by Chas. Woodson, W6NEY, 2301 Oak St., Berkeley, CA 94708. The complete 1975 and 1976 CCWN (64pp) is available for $5 each, and the 1977 sub- scription price is $10. A free subscription is offered to anyone who will build his own c.c.w. station. Also, a p.c. board and kit of parts for the Petit CCW Filter is available through Petit Logic Systems, Box 51, Oak Harbor, WA 98277. Finally, I encourage readers to familiarize them- selves with the contents of this paper at least, and at best, to begin work on a c.c.w. station. If this paper serves to motivate several of you to get into c.c.w. work, the effort expended in writing it will have been worthwhile. Amateur radio has thrived in the past on the spirit of attacking new frontiers and conquering them, and at present, c.c.w. is the frontier for that spirit. Let's rise to the occasion and once again reclaim that reputation that came from radio amateurs converting a futuristic idea.