Coherent CW for VHF will it work? Don H. Gross W3QVC RD 6, Scaite Road Sewick/ey PA 15143 Bert C. De Kat VE3DPB PO Box 137 Lynden, Ontario Canada LOR 1 To The so-called system of "coherent CW," which is actually a form of matched filtering with ex- tremely narrow band- widths, has been applied to the high-frequency spec- trum and should have inter- esting VHF applications. In- formation is typically com- municated at a bandwidth of only 10 Hz, resulting in a remarkably high signal-to- noise ratio. Successful con- tacts have been carried on at low power levels (such as 1 Watt) on 80 meters and overlongdistances(Califor- nia to Asia) on 20 meters. Interesting technical by products have been the de velopment and use of the P ' etit filter with readily ad- justable narrow band- widths, the production and use of high-quality frequen- cy synthesizer ' s, and major advances in frequency sta- bility, along with greatly im- proved methods of f requen- cY measurement. Another important factor is the use of keyers with precision timing and phasing for each bit of each Morse element, using either the Accu-Keyer or special computer keying programs. To our knowledge, th application of all of thes technologies has not bee made to VHF communica tion, but some of them ma hold promise for importan future advances at thes higher frequencies. Why Narrow Bandwidth at VHF? The use of FM and the promise of packet commu- nication of digital informa- tion at high speeds has drawn attention to the ad- vantages of the wider band- widths available at VHF. But for some applications, a completely opposite ap- proach may be better. Sup- pose that we would prefer to get maximum range or highest intelligibility for on- ly a brief message or one that might just as well go slowly. This could be, for in- stance, where the most im- portant information might be evidence of contact through call letters and a signal report. Let us make a simplifying assumption (not exactly true) that the methods of modulation and detection would be the same for ei- ther wide- or narrow-band- width communication and that the bandwidth re- quired is the same as the bits sent per second. Curi- ously, 10 kHz completely filled for one second with '10,000 bits of information (as in a packet) would trans- mit exactly the same num- ber of bits as 1,000 separate channels, each 10 Hz wide and each transmitting only 10 bits during the second (as in CCW). Noise power on each channel is proportion- ate to channel bandwidth. So, for the sam4 signal-to- noise ratio, each narrow- band signal using one milli- watt would do as well as the packet transmitter using 1 Watt. All the narrowband stations together would use 1 Watt-or if we preferred to use only one channel, we would use the same total amount of energy by taking 1,000 seconds to send the same message! Thus, neither method has an inherent advantage in bits of information per unit of transmitted energy. So our choice will be made by whether we prefer speed at higher power or slowness at lower power (or perhaps greater range at the same power, at the sacrifice of speed). We should note that if the total time is minimally used by packets or if the total frequencies are mini- mally used by lower-power narrowband signals, the chances of interference to either mode by the other mode in the same frequen- cy range are very slight. Each tends to be immune to the other. (This will not ap- ply, of course, if some greedy DXer tries a kilowatt on CCW!) The Narrowband Matched Filter We usually think of Morse code in terms of dot and dash patterns, each at- tached to a particular let- ter, number, etc. But Morse can be just as well con- ceived as a digital system based on an "on" (= one) or "off" (=zero) condition during a series of equal time intervals. Each time in- terval would be the length of a dot. A dot is a single one. A dash becomes three consecutive ones. A space within a character is usually one zero, between charac- ters is three consecutive ze- roes, and between words is seven consecutive zeroes. If the timing of the Morse transmitter is precisely con- trolled, it will be sending a serial stream of digital in- formation in classical bina- ry form. Then a receiver can be constructed with a filter and detector carefully matched to decipher th digital message. Despite the title of "co- herent" CW, there is no way to preserve the phase co- herence between the trans- mitted and received waves. Ionospheric or tropospheric media always cause some phase disturbances. The true essence of CCW is in the use of a matched filter. At code speeds used by amateurs, bandwidths of matched filters can be ex- tremely small. Typical dot lengths are a tenth of a sec- ond, producing about 12- wpm code speed. A Petit fil- ter matched to such a signal has a 3-dB bandwidth of on- ly 9 Hz. This allows for an outstandingly good signal- to-noise ratio. The Petit Filter The Petit f i [ter refers to a design by Ray Petit W7GHM. Although the de- tails of its circuit are de- scribed in the bibliography at the end of this article and will not-be repeated here, a block diagram is shown in Fig. 1. The filter has several distinct features: 1.It operates near zero beat.Usually the filter bfo is at1 kHz and it tunes to a receiver signal output very close to 1 kHz. 2. Two filter channels are used with a 90-clegree phase difference between them. This quadrature phas- ing is necessary because near zero beat there is al- ways the possibility that output in one channel alone might be in such a phase as to give almost no output. In that case, the quadrature channel output would be nearly maximum. Adding the two channels ensures an output when- ever a signal is really pres- ent. The phase shift be- tween the two 1-kHz-filter bfo signals is obtained by properly dividing 4 kHz by 4. 3. Matching is achieved by using a high-precision secondary frequency stan- dard to control all func- tions on both transmitter and receiver. This not only ensures close tolerance in receiver tuning (within 1 Hz), but also close syn- chronization between re- ceived digital pulses and fil- ter pulse sampling. A con- trolled-pulse repetition rate is not sufficient to hold this synchronization; the phas- ing of the time sampled by the filter must also be ad- justed so that the transmit- ted signal is framed within a "window" opened to each signal pulse. Ten phasing- switch positions allow ad- justment of the framing in 10-ms steps. Initial adjust- ment is made by listening for the clearest reception of a series of transmitted dots. 4. Each 100-ms window opening is the result of an integrating circuit in each filter channel. The integrat- ed output is remembered for the next 100 ms by a sample-and-hold circuit. The latter either sets the level of a tone modulator for audio readout or else crosses a threshold for digi- tal detection. At the end of the 100-ms interval, the out- put of the integrator is dumped by a shorting switch so that the next sam- ple can begin. Note that the total independence of each sampling time interval dis- allows the "ringing" so common to very selective bandpass filters. Ringing is a condition of slow buildup and decay that can make a Morse signal sound so mushy as to be unreadable. The Petit filter is immune to rin in . The selectivity curve of the Petit filter is shown in Fig. 2. It does have side lobes that are still fairly high, although they remain quite close to center fre- quency. These side lobes might be diminished or eliminated by using other kinds of filters or by modi- fying the Petit filter. They are partly the consequence of the assumption in the de- sign of the Petit type of fil- ter that the Morse digital in- formation is In pulses with zero rise and fall times-an inaccurate assumption. Bandwidth may be ad- justed easily by changing the sampling time interval. For instance, 1-second dot intervals will produce a bandwidth of only about 1 Hz, pulling signals out of the noise in a most impres- sive manner. Signal-to-Noise Improvements Some taped samples of 80-meter signals received by W3QVC and lab tests by W7GHM using the Petit filter are available for loan/purchase/copy from W3QVC at a minimal cost. On an abstract numerical basis, signal-to-noise ratio is inversely proportional to re- ceiver bandwidth. Thus, a 10-Hz-wide channel would give a signal-to-noise power advantage of 210 times (23 dB) compared to a 2100-Hz channel (a typical band- width for SSB voice com- munication). In practice, the human ears and brain allow a de- gree of concentration on 73magazine - July,1982 49 the single tone of a CW sig- nal that is the equivalent of a much narrower band- width than the wider filter would indicate alone. This effect has varying evalua- tions. One estimate (Wood- son, QST, May, 1981) is that "this skill is worth at least a 6-dB margin when using a 2300-Hz filter. QRM, how- ever, is often a confusion factor and therefore causes more degradation of copy than an equivalent amount of random noise. These psy- chological factors are diffi- cult to quantify, but prob- ably reduce the advantage of CCW over ordinary CW." Woodson then gives comparisons of CW and CCW at d if ferent power lev- els in 14-MHz communica- tions in 1975 between JR1ZZR and W61313. Both modes were taped simulta- neously on separate stereo channels and each channel was played back to four moderately experienced CW operators. The conclu- sion saw "an estimated 13-W CW signal as equiva- lent to a 0.1-W CCW signal in communications effec- tiveness, or a 24-dB superi- ority for CCW." (He should have said "21 dB" for that power gain.) The taped laboratory ex- periments of W7GHM indi- cate that a CW signal 5 dB below the white noise level using a 500-Hz filter is just barely audible and only oc- casionally readable. The addition of a 10-Hz-wide Petit filter brings it to an easily copied level at what Petit describes as a "signal- to-noise ratio of 12 to 15 dB." When Petit then drops a signal 14 dB below th noise and changes his filte to only 1-Hz bandwidth, th results are truly astonish ing. The signal goes fro completely lost in the nois up to 15 dB above th noise-a gain of 29 dB! Frequency Stability A narrowband syste can work only if its overal 5o 73Magazine 9 July,1982 frequency stability is within its bandwidth. Channels 10 Hz wide will tolerate errors of only a few Hz. Two fac- tors affect frequency stabil- ity: (1) phase changes due to variations in the propaga- tion characteristics of the medium through which the wave is sent, and (2) the ac- curacy of the frequency- control systems for the transmitter and receiver. Phase changes during propagation set an almost absolute limit to the nar- rowness of the bandwidth that can be used. How could we imagine, for in- stance, that a VHF signal broadly modulated by the undulations of aurora re- flection could be contained within a 10-Hz channel? So far we do have some experience with CCW in long-distance HF communi- cation. Woodson says: "For 14-MHz signals, motion in the F layer typically pro- duces 2 or 3'Hzof phase (or frequency) modulation for a J A to W6 path. (We have also observed what appears to be propagation time de- lays under poor band condi- tions.)" Woodson goes on to speculate about VHF ap- plications: -CCW might be used for EME communica- tion, but the problem is complicated because of lunar-motion Doppler ef- fects. One might need a computer to calculate the frequency at which the sig- nal is expected to return." A more practical solution to the Doppler problem with satellite repeaters might be reached through tight phase-locking to the satellite beacon signal, fol- lowed by computerized se- lection of the receiver fre- quency for a given transmit- ter. Even this would involve the solution of a complex puzzle. VHF experimenters will have to discover what at- mospheric conditions will allow the practical applica- tion of CCW to the VHF and UHF bands. Exactly wha phase shift is introduced in tropospheric propagation? Can frequency modulation be confined to 2 or 3 Hz? On what bands, under which circumstances? Questions like these, with answers not yet avail- able, determine the ulti- mate possible narrowing of bandwidths. But the picture is less cloudy, indeed hope- ful, when we consider the area of equipment frequen- cy control. Secondary Frequency Standards The accepted frequency accuracy for HF CCW equipment is one part in ten to the seventh power. This allows for an error of not more than 1 to 2 Hz in ei- ther the transmitter or the receiver-adequate for 10-Hz bandwidths. The re- quired precision is met by carefully constructed room-temperature oscilla- tors with temperature com- pensation through suitable capacitors across the crys- tal. VHF CCW calls for at least an order of magnitude of improvement in frequen- cy accuracy. Frequency standards dependable to one part in ten to the eighth are not so simple. They use excellent crystals and both crystal and oscillator are enclosed in two concentric proportionately tempera- ture-controlled ovens. The one ray of hope for amateur use of these standards is that they are available on the surplus market from time to time, currently cost- ing about $75. The setting of the exact frequency of such a stan- dard isalsoa problem-but not unsolvable. HF propa- gation phase shift makes WWV unusable for most people for standardizing frequency to better than one part in ten to the seventh. Higher accuracies can be obtained from one of three comparisons: (1) with WWVB at 60 kHz, (2) with Loran C at 100 kHz, or (3) with TV network color- burst signals. Comparison With Primary Standards Don Gross has developed a receiver that allows the signal from WWVB to gate a frequency counter. The' frequency of his secondary standard is multiplied' by ten, resulting in a 10-MHz wave to be counted. By us- ing 100-second gate times, his standard can be mea- sured to parts in ten to the ninth. Counting errors are typically only 1.4 digits (or 1.4 parts in ten to the ninth) during the, midday hours when 60-kHz propagation is most stable. The addition of a voltage-variable capaci- tor to the frequency stan- dard allows easy trimming adjustments to a part in ten to the ninth. Drift is so slight that such trimming is need- ed only two or three times a week. Such high accuracies are possible when WWVB is re- ceived on a good balanced and shielded loop antenna and when the receiver bandwidth is narrow enough to provide a good signal-to-noise ratio. The Gross receiver converts the 60-kHz signal to 1.11 kHz, where it passes through an N-path filter only 0.1 Hz wide. It is then re-converted to 60 kHz, limited and zero- crossing detected, then fre- quency divided to provide the counter gate control. Both the down- and the up- conversions use heterodyn- -ing frequencies derived from the secondary stan- dard. Bert De Kat has devel- oped an effective and fairly simple method of measur- ing frequency by using Lor- an C. He uses a switch-con- trolled frequency-divider system to derive from his secondary standard the pulse repetition rate (PRO of any Loran C station. (This divider is derived from Fig. 7, Burhans, 73, May, 1978, ignoring the slave window timer.) He sets his PRF to coincide with the nearest station and uses this local- ly-derived signal to trigger his oscilloscope. A broad- banded 100-kHz shielded loop and amplifier provide the Loran C signal to be dis- played on, the scope. By switching to a one-digit mis- count in the divided fre- quency, the position of the display can be slowly moved across the screen until it reaches a suitable spot; the count is then cor- rected and the waveform stays in position. By using a high-grade oscilloscope, it is possible to expand any small portion of the 100-kHz -waveform. By choosing the third zero- crossing of a pulse being built up, it is easy to keep track of the length of time that part of the wave moves across a measured part of the screen. This information can be used to measure the phase drift of the secondary standard. This measure is highly accurate, since the chosen part of the wave- form is purely ground-wave and therefore stable in its propagation. Frequency is readily measured in parts in ten to the ninth or better. Other methods of precise frequency measurement are covered in the bibliog- raphy. Frequency Synthesizers Since every frequency and timing element must be accurately controlled, high- quality frequency synthe- sizers are important. Ray Petit has done outstanding work in this direction. Al- though the bibliographic references to his synthesiz- ers do not represent his lat- est developments, they show examples of excellent equipment that can be used to tune in either 100-Hz or 10-Hz increments, all phase-locked to the second- ary frequency standard Keyers The keyer that lends it- self especially well to the timing requirements of CCW is the Accu-Keyer de- scribed in many issues of the ARRL Radio Amateur's Handbook. The oscillator part of this keyer is elimi- nated. A 10-Hz square wave derived from the frequency standard is connected in its place. This same 10 Hz is sent to the clock input of two D flip-flops: The Q out- put of one of these goes to the dot input of the keyer; the Q output of the other provides the dash input. The paddle (preferably a dual squeeze type) con- nects to the D inputs of the flip-flops. Debouncing can be arranged by connecting a resistor from each paddle output to the keyer positive voltage and a capacitor from each paddle to ground. CMOS versions of the Accu-Keyer are easily constructed and they are advantageous. The result of this circuit modification is a keyer that follows the desired CCW timing cycle to perfection. Computerized keying is becoming increasingly pop- ular. Some commercial key- ers can be modified for the external timing and phasing required for CCW; others cannot. W3QVC had hoped that his M-80 Morse pro- gram for his TRS-80 would be adaptable to CCW. Al- though its keying speed can be fine tuned, its phase can- not be linked to the second- ary frequency standard. As a result, with much that has been learned from volume 4 of 'the Disassembled Handbook for the TRS-80 mentioned in the bibliogra- phy, he is encouragingly en- gaged in the production of a machine language Morse program for the TRS-80 (ei- ther Model I or Model 111) that can use external clock- ing derived from his fre- quency standard. Conclusion CCW is just beginning to make its mark in amateur radio communication. With all the technological ad- vances now at hand, there is every reason to consider the possible. usefulness of CCW in the VHF range. There is plenty of room for new experiments! E Bibliography Burhans, R. W.: " If You Want To Know Where You Are: The Mini-L Loran-C Receiver," 73 Magazine: Part 1, April, 1978; Part 11, May, 1978. Disassembled Handbook for the TRS-80, Vol. 4, Richcraft Engi- neering, Chautauqua NY 14722, 1981. Hewlett-Packard: Timekeeping and Frequency Calibration, Ap- plication Note 52-2, 1976. National Bureau of Standards: Davies, Kenneth: Ionospher- ic Radio Propagation, NBS Monograph 80,1965. NBS Radio Stations. NBS Time and Frequency Bulletin. The New NBS Frequency Calibration Service. (On use of TV network color-burst signals.) Time and Frequency Users' Manual, N BS Technical Note 695, 1977. Petit, Raymond C., W7GHM: "Coherent CW: Amateur Ra- dio's New State of the Art?", OST, September, 1975, p. 26. "Frequency Synthesized Lo- cal Oscillator System for the High Frequency Amateur Bands," Ham Radio, Octo- ber, 1978, p. 60. "Phase-Locked 9-MHz BFO," Ham Radio, November, 1978, p. 49. "Phase-Locked Up-Convert- er," Ham Radio, November, 1979, p. 26. "Technical Topics," Radio Com- munication (RSGB), June, 1975, p. 462; July, 1976, p. 517. Weiss, Adrian, W8EEG: "Coherent CW-The CW of The Future?", CO: Part 1, June, 1977, p. 24; Part 11, July, 1977, p. 48. "ORP," CO, January, 1978, p. 44. Woodson, Charles, W6NEY. "Coherent CW," OST. Part I-"The Concept," May, 1981, p. 11; Part II-"The Practical Aspects," June, 1981, p. 18. "World of Amateur Radio," Wireless World, March, 1975. 73 Magazine e July, 1982 51 CORRECTIONS "Coherent CW for VHF," which appeared in the July, 1982, issue, was based on a paper given at the 27th Annual VHF Conference, Western Mich- igan University, Kalamazoo MI, on Oct. 17,1981. Tim Daniel NBRK 73 Magazine Staff