CFA and DLA Discoveries
Since my article last month, December 1999 Issue No. 32 of antenneX, I have done some DLA and CFA experiments and discovered the usefulness of a software program called PSPICE. This is part of a continuing effort to understand these antennas. I think I have found a really simple way to feed the CFA, and suggest that the DLA is not really a CFA without further structures, which I attempt to describe.
CFA
In the article last month, CFA: Calculating Dimensions (now in antenneX Archive III), I described the D plate capacitance and its relation to the ratio of E plate to D plate voltage. I chose a D plate size (based on the size of a sheet of brass in the basement) and computed its capacitance and E/D voltage ratio for different spacings of the D plate to the ground plate.
The next logical step seemed to be to scale the proportions of the antennas at Barnis, Egypt where we have a picture of the antenna with a man standing in front of the CFA station antenna followed by measurements of the unit described in the article on constructing a 75-80 meter antenna in antenneX. The following tables show this calculation.
To Generate Table 1a
I used vernier calipers to measure the dimensions of the antenna and man in inches on the picture. These values are shown in the column labeled “picture”. Those values are normalized into units of “men” by dividing by the size of the man. I guessed he might be 66 inches tall, so I then knew the size (within the % error in the guess) of everything in inches. No attempt was made to account for the foreshortening at high vertical angles. The operating frequency is known, so, next, everything can be transformed into size in wavelengths at the operating frequency. In particular, the D plate diameter is .0111 wavelengths in diameter. I then found what frequencies make a 12-inch and 20-inch (30 & 50cm) diameter D plate .0111 wavelengths across. These frequencies are shown along with the sizes of everything else in the “12” and “20” columns. The reactance of the two small D plates and the full sized unit at the respective operating frequencies was computed in case it was near some interesting number like 50 ohms or 377 ohms space loss.
The same process was done on the 75-meter CFA described in antenneX for comparison. Table 1b shows these calculations. Since I used a copy of table 1a’s MS Excel spread sheet again, and the CFA was defined in meters, I had to invent an imaginary man 1 meter high. It would be interesting to know these dimensions in wavelengths for other working antennas to see if there is a preferred size.
Another interesting piece of information
Another interesting piece of information is the effective capacitance of the E plate, and the way it varies with height above the D plate. This was done by using the E late capacitance in a series-tuned circuit with a known inductor (it doesn’t matter too much what value). The voltage dip at resonance is easily seen at the signal generator’s output cable when this tuned circuit is put across the cable to ground. Table 2 shows this experiment. The first column is the resonant frequency for the described conditions. The reference capacitance is the D plate, which is calculated to be 62 pF. This gives the coil size, subject to errors in the D plate capacitance. The testing L is 69 uH. The second column gives the height of the E plate cylinder above the D plate for a number of heights with the D plate shorted and again with it open. The resonant frequency and “known” inductor computes to the capacitance values in the last column. I was surprised to find that the grounding of the D plate had so little effect; I think it means that few field lines from the E plate terminate on the D plate. This supports the assumption that these lines describe a quarter circle leaving the cylinder horizontally and ending up on the ground plate. The last entry is the capacitance of the net of clip leads connecting everything together without connection to the E plate.
About the time the above was done, I decided that doing the network calculations for CFA feeds by hand was too time consuming. After all, computers should do-so that people can think. I hunted around and found a site that offered student evaluation versions of this neat program found at http://www.repairfaq.org/ELE/F_Free_Spice1.html One doesn’t need the model for some advanced microchip; all that was necessary was the R, L, and C models plus transformers and coupled inductors. I strongly recommend getting a copy of spice! All the following calculations were done using Microsim’s Version 8 for Windows 95. At this site are all kinds of versions, for all kinds of older computers. Put your boat anchor pc to work!
Also, about this time, I built a DLA (aka, Double-Loop and Dual-Loop Antennas) loop. Other experimenters had reported that the twisted wire versions of the DLA gave higher radiated field strengths, so my loop is made of 9 feet (2.7m) of twisted #14 house wire inside a “hula hoop”. I tried the inventors’ configuration first, and did many “spice” models with the wires connected in all combinations of coil interaction. Eventually I figured out that the energy goes through both coils in series rather than through each wire in parallel. I got it to work, but it was always very high Q and irritatingly narrow band. I made one 300-mile (484 Km) contact on 40 meters, but at that time the feed line was hot and probably was radiating. Finally, I wired it with the wires in series with only one capacitor for tuning. This version had acceptable SWR and somewhat wider bandwidth, but it never resulted in a contact. I then recognized that it could be connected as a 1:1 current balun. Spice told me the interesting fact that at resonance, the two balanced wires had equal voltages and phases of + and – 90 degrees! Just the kind of thing to drive a CFA! The result of further spice modeling including the D plate capacitance is the second of the feed schemes described below.
While modeling the CFA feed Schemes
I wondered if it made any important difference if all the reactance values were low – like 50 ohms, or 377 ohms, instead of whatever the D plate impedance was. All the other networks described in antenneX used additional tuning capacitors in parallel with the D plate. In all my models, I simulated the transmitter and its cable with a 1-volt source and 50 ohm series resistor. A perfect match happened in the model when the network loaded this transmitter with a perfect 50 ohm resistor, and the voltage at the cable/network interface was 0.5 v with 0 degrees phase angle. After trying various tuning capacitors in parallel with the D plate, I found that one can choose a value that gives 50 ohms load with only a small phase angle, about 15 degrees. The E plate voltage could be shifted 15 degrees with a large, low reactance series capacitor. The result was the model in Figure 1, which computed to the curves for load voltage at the cable (node 2 on figure 1) and the D plate voltage (node 3) shown in Figure 2. The space loss at each plate was assumed to be 377 ohms.
Figure 1: Matching network using partial loading of D plate capacitance by space loss
Figure 2: Tuning curves for transmission line load and D plate voltage in Figure 1
When I Tried it in the Lab
When I tried it in the lab, though, the large series capacitor (C3) had no visible effect! Which led to the question: Where’s the load?? All the networks I have tried had high Q, low loss properties. I believe that radiation loss should show up in the experiment as a reduction of Q somewhere. Radiation is, after all, just another form of dissipation of energy. It doesn’t matter how it leaves the system. If the E plate doesn’t radiate, then two effects are visible: First, R3 in Figure 1 doesn’t exist, so C3 in Figure 1 is connected to nothing except the E plate capacitance of about 12 pF, so adjusting it has no effect. Second, reconnecting the adjustment capacitor in a parallel resonant circuit excited by the D plate will give a narrow band, high Q adjustment. When this condition, Figure 2, was tried, it behaved exactly that way. Adjusting the resonance greatly changed the matching, and the whole antenna became sensitive to touching, etc., but it was very narrow band and the best setting was hard to find and hold. Changing the inductance and adjusting C accordingly, to see the effect of L/C ratio, didn’t change anything noticeably. I conclude that the E plate doesn’t radiate, at least for the configuration of Figure 3.
Figure 3: Workable E plate tuning which still effects D plate
Figure 4: Balun Fed CFA
Figure 4 above shows the best network I have found – the balun feed. It is easily adjusted with only one tuning capacitor. The bandwidth is large, 6.916mhz to 7.552mhz between 2:1 points, with 1:1 at 7.285 MHz.. I think that I can tune this thing up to 20 meters with one tuning capacitor! Table 3 shows the component values for the balun and tuning C. The curve for the computed tuning capacitance, 286 pF, is shown in Figure 5. It shows a match at too low a frequency; adjusting this value to 241 pF gives curves that agree with the matching points described above and in Figure 6. The curves show D plate (v(2)), E plate (v(3)), and ground plate (v(4)). This network results in an antenna that acts like it is radiating – A clip lead on my RF probe receives the milliwatt from the signal generator at 3-4 feet, something no other experiment has done.
The Next Experiments
I welcome any dialog about these subjects, so please try to duplicate the above and send me your experiences.
Originally posted on the AntennaX Online Magazine by Joel Hungerford, KB1EGI
Last Updated : 2nd May 2024