CFA Tuning for Radiation
Last month I discovered that my D plate had considerable self-inductance, which combined with the D plate capacitance to produce a series resonance in addition to all resonances from the tuner. This month, I started to track this property down, but got distracted by two interesting related subjects.
Heikki Antman, OH2BGC (Finland) reported good results with a simple 3 wire CFA on 20 meters that used the balun tuner and a 5 meter length of house wire. Since 5 meters is close to a quarter wave at the 20-meter band, I wondered if this length was important. I measured this type of antenna for two lengths and several tuner configurations. The results are shown in Tables 1, 2, 3, 4,and 5, and the associated Figures.
In the middle of the night, while puzzling over the D plate inductance, I woke up with the following idea: If about a quarter wavelength of TV twinlead antenna wire is wound in a flat coil (like a watch spring) the two wires would each form a flat spiral plate of a capacitor or D plate. This plate, if fed from the center of the coil, would have an E field phase change radically due to the quarter wavelength along the spiral from feed point to outer circumference. Magnetic fields from displacement current at the center should then be in phase with E field at the circumference – a self-phasing D plate. I decided to try it, first with an available length of twinlead, and then with that twinlead cut to ham band length based on the results of the experiment with the longer length. These results are shown in Tables 6 and 7.
3 Wire CFA Tuner
When I heard about the 3 wire CFA working, I grabbed a meter stick and hunted through my selection of wire bits until I found one just 6 meters long. The experiment was to rig a tuner with clip leads and two 500 pF capacitors and feed the house wire antenna to see how it behaved. The balun was constructed by winding 11 turns of two insulated #14 wires (from taken-apart house wire) side by side as if they were one wire around a 2-inch PVC pipe 3 inches long. The ends of the winding go through two holes at each end of the tube. This gives 5 uH inductance for each winding. A hot melt glue gun fixes the spacing and position of the wires with 4 glue tracks down the coil.
The data is the frequency range over which this length of 3 wire antenna can be tuned to VSWR of 2:1 or less, and the bandwidth between the 2:1 frequencies at each setting. The MFJ antenna analyzer was used for all tests.
Table 1 shows the results of the first test, with the long antenna. The first column is the frequency. The analyzer was tuned to this number and C1 and C2 were adjusted for best SWR. It’s an easy two-handed adjustment rather like balancing a Wheatstone bridge. The 4th column is the resulting best SWR. Then the analyzer was tuned to find the frequencies above and below the adjustment number where the SWR rose to 2:1. The 5th and 6th columns give these numbers. The next two columns indicate when C1 or C2 is fully open (minimum capacitance) or closed (maximum capacitance). The last column gives the antenna length in wavelengths at the frequency in column 1. Back to the other side of the table, the second column gives the length of a quarter wavelength at this frequency, and the 3rd column gives the bandwidth between 2:1 SWR frequencies.
After taking Table 1 data, one meter of wire was cut off the antenna and the data of Table 2 was taken. All later tables are at the new length. The first thing one notices comparing Tables 1 and 2 is the antenna length shown as 5.48 meters and 4.57 meters. This is because I found the next day that I had grabbed a “yardstick” by mistake instead of a “meterstick”. Oh well – if NASA can do it, so can I!
The data shows that a longer wire lets the capacitor C1 (see Figure 1 and Figure 2 for the circuit) tune the balun to a lower frequency at maximum capacitance, but doesn’t affect the upper frequency behavior much. It is C1 that is the limit because C2 is only about 1/2 meshed when C1 is fully meshed at the low end of the band. The region of good SWR numbers is approximately centered on the frequency where the antenna is a quarter of a wave long. There appears to be another region of low SWR above 17 MHz, which widens the distance between 2:1 frequencies until about 20 MHz, where the SWR rises sharply.
I was not sure whether the input and output grounds were supposed to be connected, so Table 3 shows the tuning range if the grounds are connected (Tables 1 & 2 have separated grounds). Connecting the grounds seems to eliminate the extra resonance mentioned above, and allows a very wide tuning range.
At this point, all the clip leads and wiggly connections were removed and smaller capacitors (12-256 pF) substituted for the big 500 pF transmitting capacitors used in the first experiments. Everything was mounted in a Lucite box, and 3 banana terminals mounted to connect to the antenna wire. That wire was inserted inside a 15-foot long PVC pipe, which holds the 3-wire part vertical. One QSO was had with a station in Los Angles with the antenna connected with separated grounds, as in Table 2.
Later e-mails established that Heikki’s configuration has the grounds connected, and C1 goes to ground instead of to the other balanced output of the balun, and that the whole system is easier to tune by only leaving C2 in the circuit. He also talked of QSOs on 80 meters with the same set up. So I brought it all in out of the rain and took the data in Tables 4 and 5. Table 4 is essentially Table 3 but referring to the new, mechanically stable configuration. Using C2 only to tune the system greatly restricts the range of operation and the SWR is much worse. The extra resonance is back, giving very much wider bandwidth at 16 and 17 MHz.
I was pleasantly surprised to find the system could be tuned to the lower part of 80 meters, too. Table 4 shows that data, taken by adjusting the analyzer frequency and re-tuning. On 80 meters, C1 had no effect except to raise the best SWR number, from values like 1.3:1 up to 1.9:1 at each frequency. On 80 meters, with this balun, only a single capacitor is needed. Perhaps the balun is too big on 20 meters???
Table 5 data was taken to compare the performance of the first balun feed I tried months ago with the CFA, where one wire is connected to the input of the balun on the center conductor of the cable to the transmitter. Figure 3 shows the circuit. This system has a much smaller tuning range than the other configurations, and needs both capacitors in the circuit to get really good SWR numbers.
A last experiment was done, which needed no tables to discuss. I wondered if feeding one of the 3 wires in the opposite direction would lead to any CFA action as the signals passed each other in the antenna. Feeding from one end means everything goes down the 3 wires together, at the same velocity, and so the correct phasing for a CFA must be achieved at the tuner. So the top end of the 3-wire antenna was brought around to the input and the black wire was fed from the opposite end. The short answer is that there were no low SWR frequencies below 46 MHz! Reversing the direction of feed is not a useful option. Remember, this is not like a DLA antenna, since the non-fed end of the 3-wire antenna is left open, like a quarter wave vertical.
The Coiled Twinlead Device
I call it a device since I do not know yet whether it will actually radiate. It certainly couples to everything in its immediate vicinity – the wave of a hand near it changes the SWR achieved from 1:1 to 10:1! But lots of work needs to be done on the method of feeding it, etc. Table 6 shows the performance of a coiled 31.5-foot long length of twinlead over a frequency range from 2 to 14 MHz. The data was taken two ways: with a HP606 signal generator feeding the center two wires of the coil and the input voltage monitored by an RF voltmeter, and by recording the resistive and reactive components at the same frequencies as measured by the analyzer. To speed things up, I hooked a digital voltmeter to the output voltage of the RF voltmeter and read that, as the needle of the RF voltmeter is too far away on the bench to read without pushing the antenna tuner off the bench and upsetting everything. So, in Table 6, the first column is the frequency, the second column is the digital voltmeter output, and the third column is the digital voltmeter readings scaled to 1 volt for the largest voltage found.
The 4th and 5th columns are the resistive and reactive components of the impedance presented by the coiled twinlead to the antenna analyzer. The next two columns convert this to an impedance magnitude and phase angle.
Looking at the data shows that from 2 to 5 MHz, the impedance drops with frequency and the phase angle is near 90 degrees. The coil looks like a capacitor, and the value of the capacitance that has the measured reactance is rising rapidly as frequency increases. At 5.2 MHz it turns inductive, the impedance rises until 5.5 MHz, when it turns capacitive again for 2 data points. Then it turns inductive from 5.6 MHz to 6 MHz, characterized by a rising impedance with frequency. The value of the effective L that makes the measured reactance drops with frequency.
At 6.3 MHz, the 31.5 feet of twinlead is a quarter of a wavelength long and the impedance is beyond the capabilities of the analyzer to quantify. The calculations for this point are shown above the table where the twinlead properties are listed including velocity of propagation in the twinlead.
At 7 MHz, above the quarter wave resonance, the coil becomes capacitive again, with a decreasing reactance value with frequency and an increasing size of the effective capacitor. This complicated behavior is shown in the last 3 columns.
In other words, at frequencies below that frequency where the twinlead in the coil is a quarter wave long, that coil is short and the phase gradient from the center to the outer circumference is less than 90 degrees. As it nears 90 degrees, some kind of complicated interaction occurs, and the coil shows several points where the impedance is 50 ohms and matched. At frequencies above the quarter wave point, the coil is well behaved, unmatched to 50 ohms, and relatively un-interesting.
In the interaction band of frequencies, around 5.1-5.2 and 5.6 MHz, the coil lights up, causes deflection on the field strength meter (a crystal set with a meter on it) and is very sensitive to its surroundings. A great candidate for an antenna! But it is not in any band we can use to radiate – it must be scaled a little bit up to work at 7 MHz to be tested in the 40-meter band.
Table 7 is that scaling calculation. Table 7 shows all the frequency points from Table 6, and shows the wavelength, the electrical length of the 31.5-foot coil at each frequency, and the two-way degrees of phase delay the coil twinlead has at that frequency. The last two columns in Table 7 show the length at 7 MHz that duplicates the coil behavior found at each frequency and the amount that would have to be cut off (- delta numbers) or added to adjust the coil to the state measured at each frequency. I cut about 8.5 feet off the coil in order to try it at 7 MHz. With a clip-leaded together tuner, it showed the expected sensitivities at 7 MHz, loaded up with a low SWR, and coupled to everything around. I could faintly smell ozone after running a few minutes. It is interesting that 0.2 wavelength, a length in the interaction band, is the element spacing in a 3 element Yagi.
Originally posted on the AntennaX Online Magazine by Joel C. Hungerford, KB1EGI
Last Updated : 22nd May 2024