Coherent CVV The Concept Part 1: Would you think that you could decrease your transmitter output power by a factor of 10 and increase signal readability by the same amount - simultaneously? It's being done now. By Charles Woodson,* W6NEY simultaneously? It's The more we know about something we seek, the easier it is to find. This prin- ciple applied to Morse cw- communica- tions is called coherent cw or ccw. On-the- air trials of this technique have shown it @N ill providi: an improvement of more than 20 dB in communications effectiveness o%cr ordinary cw methods. This same principle can be used with RTTY. ASCI I and fisk signals. but this discussion will focus on cw keying. Cw signals may be analyzed as a series of digital units. all of which havc (at least approximately) a unit of time in common. For convenience. I'll call this time unit a "frame." Each frame contains either a ..mark" (key down) or a "space" (key up). Fig. I illustrates this concept. Ordinary cw dots, dashes and spaces begin at somewhat arbitrary times, depending on when the operator happens to press.the key. Thus. the frame length varies to a considerable dcgrcc, and you can't predict when each framc starts and ends. With ccw, all dots, dashes and spaces arc exact multiples of the basic timc unit and occur within predictable timc frames. This includes any pauses during transmission. When received, ccw signals sound like any other cw signal cxccpt that they arc being sent very precisely, as with P1q. 1 - The elements of ccw communication. Frames. in 0.1-second units. are shown on the horizontal axis. The enable (top waveform) shows the I losure of a manual key by the operator. When referenced to the precise frame times. it can be seen that the dots. dashes and spaces of the enable ire not accurate in length. Note that with the ccw-keyer waveform a mark or space is begun only at the beginning of the frame period and continues 'Or the full period(s). As received. the Signal is mixed with ORM and ORN. The difference between the dc voltages from the switching mixers of the Ivio channels (third wavefo(m) is a junction of the desired. but weak. signal. An integrator sums the power (voltage) received over the frame period. This sum is sampled at the end of the neriod and held until the beginning of the next period. The recovered modulation is used to key an audio ,,ignal for detection by car. May 1981 11 ENABLE CCW KEYING TWO-CHANNEL DC-LEVEL DIFFERENCE TWO-CHANNEL INTEGRATOR DIFFERENCE TWO-CHANNEL HOLD DIFFERENCE RECONSTRUCTED MODULATION @ 2301 Oak St.. Berkeley. CA 94708 FRAME TIME IN 0.1-SECOND UNITS a perfect "fist." This characteristic is utilized to permit the use of vcry narrow bandwidth Filters. CW Filters In general, receiver Filters with band- widths much widcr than that of the desired signal arc less effective because they allow reception of additional noise and undesired signals. At 12 wpm a cw signal occupies about 10 Hz of the spec- trurn, yet 500- or 2300-H/-wide filters are frequently used for cw reception. With a 500-Hz filter, one hears the 10-H/-widc desired signal and 490 Hz of noise and QRM! By analogy. an ssb operator using a similar approach would listen to IM kHz of the band at one time! High-Q analog cw filters are not useful at the narrow bandwidths approaching the bandwidth of a 12-wpin cw signal. Such filters. with bandwidths less than 5M Hz, tend to "ring" or produce an output after the signal ceases. The human car is confused by such ringing. Also, the receiver stability and resetability required in conjunction with the filter. on the order of a few hertz. is difficult to achieve. 11hase-locked loop (131-L) filters with time constants long enough to produce bandwidths of only a few hertz unfor- tunately take tens of seconds to achieve lock. 111-1-s also tend to lock on the strongest signal in the passband and are. therefore, sensitive to QRM. I'LL filters have their place of importance. but not with the bandw 'idths required hcrc. The filter we need will providc a band- width of only a few hertz without ringing and without a tendency to lock on the QRM. Such a filter improves the sighal- to-noise ratio dramatically. A I-W signal copied through a IOL-Hz bandwidth Filter is comparable to a 50-W signal heard through a 5(X)-Hz filter or a 230-W signal heard through a 2300-Hz filter. The CCW Station Typically, ccw stations agree on an operating frequency (e.g., 14,049.(XX) Hz ñ 2 Hz) and a frame length (usually 0. I second, the speed of 12 wpm). and acquire the "training" - when each frame starts and ends - as part of the signal-tuning process. Thus, the frequency, frame length and franic phase arc all known at the receiving end and are used to advan- tage in the detection process%. To achieve the necessary franic-Icngth accuracy and to get on the operating fre- quency within the narrow tolerance of the filter ' all frequency-determining oscillators in both the transmitter and receiver of the ccw station must be highly stable and accurate. The stability and ac- curacy requirements are obtainable by using carefully built crystal oscillators which arc compared to a reference fre- qucncy such as WWV. Time discipline for the transmitted signal is determined by a reference oscillator which is divided to provide a 10-H/ synchronizing signal for the transmitter keycr. The ccw filter at i lie receiving station uses timing signals de- rived from the station reference oscillator. These timing signals tell the rccci%er filler when to expect a frame to begin and clid. The Coherent Initegrating Filter Fig. 2 shows a block diagram of' the filter which makes possible the efficient reception of a ccw signal. The nia - ior blocks of each of the two filter chains ai c: input mixers, integrators, sainpIc-and hold circuits. output mixers and 111C timing and control circuitry. The reason for the two chains will be examined later: for now, we'll follow the signal through one chain. The Alirer: The first part of' each filter chain is a switching inixer %%-here tile desired signal (along with adjacent QRN and QRM) is mixed with a refercricesignal of the same frequency as the LICSIrcd signal. (Solid-state switching is perf - orim'd in the actual circuit. but for siniplicil\. mechanical contacts are shown ill I ig. 2.) The reference signal is obtained front a stable source such as the timing and Coll- trol circuitry. and it determines the center point of the ccw filter. A signal at the desired frequency conics out ofthe Illixer as a de voltage - the stronger the SiCnal. the larger the %oltage. Ali oft-fircquelI0 signal, however, conics out ol'the lin\cr a, a low-fircquency ac voltage. We mi\ OIC incoming signal right down to /cro I)CZII- Undesired signals will be distinatiishcd INPUT + 4 NETWORK CENTER FREQUENCY X4 REFERENCE CHANNEL A SWITCHING MIXERS INTEGRATORS SAMPLE OUTPUT MIXERS I DUMP HOLD SOAI 1Z C CIA SSA SMA2 CIA 1Z DA Smal SM82 CHANNEL B Sse SMAJ SMEIII SMA2 SMS 2 is DB 1Z DUMP AT END SAMPLE JUST OF WINDOW BEFORE END OF WINDOW COHERENT - INTEGRATION SAMPLE / DUMP CONTROL CHA CHB PHASE ADJUSTMENT SOA2 SOBI SOB2 + 4 NETWORK DESIRED OUTPUT FREQUENCY X4 COMBINER -0 AUDIO OUTPUT COHERENT - WINDOW REFERENCE SOAI SOBI SOA2 SOB2 Fig. 2 - Block diagram of a ccw filter. 12 UST_ from the desired signal because they arc not cxactlv /cro beat. The Integrator: An op-amp integrator comprises the second part of each Filter chain. We use the integrator to distinguish the desired signal (the /.cro-beat dc @oltage) from the undesired signals (low- frcqucncy ac voltages) coming from the mixers The integrator may be thought of as a moderately large capacitor. A syn- chroniAng "dump" signal from the timing and control circuitry shorts out this capacitor at the start of each time frame. Any desired signal (de voltage) during the tinic frame causes the capacitor to charge. Tile resulting voltage at the end of the time frame is a function of the strength of the desired signal received during that frame. QRM and QRN, being off frequency, appear as ac signals to the integrator capacitor. These charge the capacitor for part of the time frame. but discharge it for other parts of the same period. Conse- qucntly, signals off frequency do not have as great an effect on the integrator output as do signals exactly on the desired fre- qucncy. That is how the ccw Filter achieves its selectivity. As an example, consider an interfering carrier which is 10 Hz above or below the desired signal. Following the switching mixer, this QRM appears as a 10-Hz ac voltage. If the filter is set to the ccw stan- dard frame length of 0. I second, then the 10-Hz interfering signal goes through one complete cycle during the integrating period. At the end of the time frame, the QRM-produccd voltage at the integrator output is zero. Thus, the ccw filter has a null just 10 Hz above and below its center frequency. There arc also similar nulls at other 10-Hz multiples. Sample-and-Hold and Integrator Reset: At the end of each time frame, a "sample" signal from the timing and con- trol circuit transfers the voltage at the in- tegrator output to the sample-and-hold circuit. That circuit "remembers" that voltage for the following interval. Once the sampIc-and-hold has acquired the in- tegrator output voltage, a dump signal from the timing and control circuitry shorts out the integrator capacitor. It does this by means of a cmos analog switch connected across the capacitor. This allows the integrator to start over again with zero voltage at the start of the next time frame. Rcselting the integrator at the end of each time frame lets tile ccw filler avoid the ringing (or intersymbol interference) common to other narrow-bandwidth Filters. Note that this is possible only because the ccw Filter "knows" when each time frame begins and ends. It is here that the time discipline of the transmitted signal is used to advantage in the detection process. Output Mixer: This last block of the filter chain is much like the input mixer: it functions as an amplitude modulator, using the sample-and-hold output voltage to control the amplitude of a sidetonc. The purpose of this mixer ii to construct a sidctone for the human operator to hear. Why Two Channels? If the incoming signal is in phase with the center reference, then the mixer out- put is always positive. The integrator which follows will see a positive de voltage. If the signal is out of phase with the reference, then. the mixer output is always negative. The integrator will see a negative de voltage. The positive or negative devoltage charges the integrator capacitor, the sampic-and-hold '.remembers" that charge during the next time frame. and the output mixer generates a sidetonc whose amplitude is proportional to the voltage on the sample- and-hold capacitor. But if the signal is 90' out of phase with the reference frames, then the mixer output will at times be positive and at other times be negative during a given input cycle. This output will be averaged to zero by the integrator. The result is no filter output from this channel. The situation is different for each chan- nel because the A channel input mixer is operated by a reference which is 90' out of phase with tile B channel reference. Thus, if a signal is 90' out of phase with the A channel. it will be in phase (or 180' out of phase) with the B channel. At all phase differences between the two chan- nels, the product of the two channels is always the desired signal despite the phase relationship between the center frequency reference and the incoming signal. If the desired signal is graphed as a phasor (as in Fig. 3) one might say that the It channel picks up the X component of that phasor, and the A channel picks up the Y component of the phasor. The two- channel output mixers arc also driven with signals 90' out of phase. That way, the output tones combine vcctorially. The result is that the combined output is a tone whose amplitude reflects the amplitude of the desired signal. regardless of the signal phase. The phase of the output tonc also reflects the phase of the desired signal. The theoretical response curve of the filter may be developed. We won't go into the mathematical details except to say that the amplitude response is a sin x/x curve, like that in Fig. 4. For a 0. I -second frame length, the nulls in the filter response oc- cur every 10 Hz either side of the center frequency. The 3-dB points on this curve arc 9 Hz apart; the 6-dB points arc 12 Hz apart.' Fig. 5 compares the ccw filter (0.1 see- ond frames) with an ordinary 500-Hz cw filter and a 2700-14z ssb filter. On this scale it is impractical to show the numerous nulls in the ccw-fiItcr response; shown instead is the envelope of the primary response. How Much Improvement? One way of comparing ccw with the or- dinary cw method is to consider the filter noise bandwidth. This is the bandwidth of 1-- a W UJ C.) W cc Z W Z R 2 # 0 1 C.) I Z -i X 9 - - - - 30 S-CHAIN COMPONENT RECEIVED L Fig. 3 - The desired signal considered as a phasor. I I -17dS PEAK -17dS I EAK AT - 20 Hz NULL ULL AT AT +2OHz -10 10 Hz /I,",- -2OHz -iOH2 CENTER FREQUENCY Fig. 4 - Filter-response curve for a 10-Hz bandwidth ccw filter. +l0HZ +20HZ May 1981 13 an ideal stccp-sidcd filter which would pass the same amount of random noise as the filter being considered. For 0. I -second frame length ccw, the Filter noise band- width is 10 Hz. This equates to- an ap- proximate superiority of 17 dB over a 5(X)-Hz cw filter and about 24 dB. over a 2300-Hz filter.. Such estimates should be reasonably accurate with respect to noise. but when QRM- is present, the ccw Filter probably does even better. Using a ccw system of 0. I -second frames with ground wave in the presence of natural noise, and adjusting power for matching readability, I have measured an approximate 16-dB improvement over. a 470-Hz crystal Filter; this is close to the theoretically expected value. Narrowing the ccw bandwidth by using longer frame times provides an additional signal-to-noise advantage at the price of slower information transmission rates. A 0.1-sccond integration period gives about 24 dB improvement over a 23(X)-H/. crystal Filter: a I-second integration period (1.2 wpni), 34 dB; a 10-second period. (0.12 wpni). about 44 MI. These speeds. are slow, but the improvement in effective communication with lower power is quite fascinating. The improvement gained by long-franic ccw is limited by phase modulation in- troduccd by the propagation path. For 14-MH/ signals. motion in the F layer typically produces 2 or 3 Hz of phase (or frequency) modulation for a JA to W6 path.' (We have also observed what ap- pears to be propagation time delays under poor band conditions.) When the Filter passband becomes so narrow that this modulation exceeds the filter bandwidth, further improvement in signal-to-noise ratio cannot be obtained by narrowing the filter passband. In evaluating Filter effectiveness, noise bandwidth does not tell the whole story; there are psychological considerations. too. The human car is frequency sensitive, and the human brain can focus on par- ticular cw signal frequencies amid the noise and QRM. Skillful cw operators use this capability well. My observations have led me to conclude that this skill is worth at least a 6-dB margin when using a 23(X)-Hz filter. QRM. however, is often a confusion factor and therefore causes more degradation of copy than an equivalent amount of random noise. These psychological factors arc difficult to quantify, but probably reduce the ad- vantage of ccw over ordinary cw. Fig. 6 shows graphically the results of on-the-air comparisons bctwcen cw and ccw made in 1975. Transmissions were made on 14,049,000 Hz from IRIZZR at power levels of 10 watts, I watt and 0.1 watt using ccw and a vertical ground- plane antenna on a four-story building. A th.rec-clenient beam was used for reccp- tion at W61313. The ccw signals were received simultaneously as cw and ccw signals, and were recorded on separate channels of a stereo cassette recorder. We selected sample periods from the cassette recording and played back the signals to four moderately experienced cw operators. The average proportion of' copy shown on the graphs is based upon words Lonsidcred copied. The copy con- Concluding Remarks The ccw technique appears to be most promising, especially where signals arc weak compared to the noise and QRM. Under high absorption and QRN condi- tions (as often experienced on 80 and 160 mctcrs) the additional selectivity of ccw would be hclpfu4 we don't have data on that yet however. Ccw might be used for EME com- munication. but the problem is com- plicated because of lunar-motion Doppler effects. One might need a computer to calculate the frequency at which the signal is expected to return. Also. achieving the necessary frequency stability of I or 2 Hz is more difficult at the higher frequencies used for EME. Some of the simplest rigs arc the easiest to convert for ccw operation. To obtain the full advantage of the ccw mode. however. receiver quality should be high. In Part 2. 1 will describe the equipment and methods used for communicating by ccW. Oif- I - to co - 20 LU V) Z 0 W M - 30 IX 0: W 1"; -40 i: - 50 -60 r TYPICAL CcW FILTER 2700-Hz 10-BAUD, tO-Hr Ssa FILTER 11ANDWIDTH I I rYPICAL 500-Hr I CW FILTER --7 - r- - 500 Hz 0 + 500 Ns FREQUENCY Fig. 5 - A comparison of three filter-response curves. .[Editor's Note: The amount of "Signal spreading" is determined in farge measure by the earth's geomagnelic activity (A-inclex). which is more isevere under disturbed condi- tions.1 14 UST@ 1.00 0.50 low IW 0.1 W Fig. 6 - A graph of the average proportion of copy made by four operators of simultaneously sent cw and ccw signals. Three different power levels were used. See text, tent was taken from radio journals. Ex- trapolation of these data indicate an estimated 13-W cw signal as equivalent to a 0. I -W ccw signal in communications cf- fectivcncss. or a 24-dO superiority for ccw. CM. References Petit. "Coherent CW: ArTrateur Radio'% Nc%, State Of The Art?". QST. Septoriber 1975. Sckiiie. "Cohercrit CW %% a Nande%uka (%k hat is ('ohcreni Japanese Hany Radio Journal, January. 1976. Weiss, "Cohereris M - The CW Of The I uturc." CQ. June and July. 1977. Iletit. "I undamental% of CM." CCW Ne"sletter 75:7. Note: Back copic% of' Volume% of the Co- herent CW Newsletter (CCWN) arc a%ailable fron, CMN. 2301 Oak St.. Rcrkelev. C% 94708: 197@. S5.1976. S5, 1977. SI(; 1971f. Sm. Volume, 75 and76 are well %unimari/ed in the Wei%% article IIICQ. Most of %olisivies 77 and 79 arc mari/Ld in this article. I urther %olunic% W the CCWN are not Planned. but a book oil :,:%% is beine assembled by Petit. This article has 11cil- fitcd troll, %u&wstions by: Arn Maynard. K-Isk; Ray Petit. W761M. K6taro.SLkinc.'JAI BlA -. Intel I'd Johnson. W2ZWA/JAlYVW.