The CFA - Frequency Verses Size
Last month (September 2000 Issue) in this series of articles on the CFA in search of radiation, I measured the “top hat” CFA at frequencies between 4 and 6 MHz. The CFA is smaller than recommended at these frequencies, so this month I measured the device at 14 MHz. I found the same behavior—that the radiation acted like the vector sum of E plate radiation and D plate radiation, and canceled when the D plate current phase was 180 degrees relative to the E plate voltage.
Sparks Fly
Part way through the measurements, the high voltage in the HP 181A/AR spectrum analyzer main frame started to arc across a bit of printed wiring, and an alternate method of remote sensing the radiation had to be devised. Before the analyzer failed, I had observed the null that occurs when D plate and E plate radiation cancel. It was easily observed by setting the screen to view from 12 to 16 MHz and tuning the signal generator across that band.
After several different attempts with passive field strength meters that were not sensitive enough, I set up a Hammerlund HQ-145 receiver as a sensor and fed the demodulated 400-Hz audio (modulation on the signal generator output) to the ac VTVM. The system was set up to give a signal about 18 dB above minimum receiver level for the smallest signals fed to the CFA. A step attenuator in the receiver antenna line was adjusted to bring the receiver output voltage to that level for all test points. The attenuator value is the data—it represents the amount the actual power must be reduced to be equal to the weakest signal found. The gain of the receiver was assumed to be the same at all points at a fixed frequency. Figure 1 shows the block diagram.
The principal difference observed between 5 MHz and 14 MHz operation was the size of the E plate coupling to the rest of the antenna and the sensitivity of all measurements to ground location. The E plate coupling showed up as a phase change in the measured D plate current. The E plate capacitive coupling to the D plate produced current in the connecting wire between the terminated 50 ohm cable and the D plate. The D plate current sensor was in that wire. The D plate impedance is about 200 ohms at 14 MHz, so 80 % of the coupled current will go through the 50-ohm cable terminating resistor and the sensor. The sensitivity of the current phase to E plate voltage level made it hard to adjust the delay for an accurate current phasing.
To design this problem out of the experiment, I disassembled the CFA and inserted a second identical round D plate under the normal one. This second “ground” plate was supported by the Lucite sheet used to make a high-capacitance D plate (all the various holes matched, etc.). It was connected to ground by a short wire through the current sensor toroid. This change greatly reduced the measured change in D plate current phase caused by the E plate coupling. The delay could then be set up using maximum D plate drive and a small E plate voltage. Since the null in radiation that happens when E plate and D plate oppose each other and are equal in radiation occurs with the E plate drive 10 dB down from the maximum at maximum D plate drive, (see Table 2) I concluded that any improper setting of delay would be small for any drive combination to the two plates that generate appreciable D plate radiation relative to that of the E plate. The current phase still changed with position of the ground wire due to currents in the ground sheet. It seemed better to then set the delay using the phase between D plate and E plate voltage (=90 degrees). These difficulties in phasing due to plate interaction and ground sheet currents increased noticeably with CFA size in wavelengths and can be expected to be further increased by circulating currents from any resonances in the tuner!
Table 1 shows the detected radiation at various combinations of E plate and D plate drive, with both plates radiating in phase and adding. The numbers are the dB setting that reduced the detected audio to the reference level for each combination. The receiver is operating at the same signal condition for each data point. Note the characteristic behavior: the level changes along any row or column until the changing drive is small compared to the other plate’s fixed drive.
Table 2 shows the detected radiation with delay cables swapped to make the radiation from each plate 180 degrees out of phase and canceling when the amplitudes are also equal. The frequency is slightly different between the two tables, by a few hundred kHz, as indicated by the larger dB numbers (more on that later). The frequency adjustment was needed because the cable combinations possible by swapping a cable to the other plate and adjusting delay using little sections of cable did not make exactly the same number of degrees delay in each side. Note this table shows the characteristic null that occurs when the E plate drive is 10 dB down from the D plate drive (shaded areas) with signal rising on each side of the null. The null also correctly tracks the change in both levels downward as drive is jointly decreased in both plates. This indicates to me that the null is not due to cancellation with some third signal path around the CFA.
There are No Miracles
Next, I set out to generate the swept frequency numbers to show the null in radiation versus frequency that is so easy to see on a spectrum analyzer. Table 3 shows the results. It illustrates how one needs to be always very suspicious and cross check an experiment every way possible. This table shows, as Andy Przedpelski, a previous boss of mine used to say—”There are no miracles!! The experiment is doing exactly what it was designed to do, which is not always what you designed it to do!!!”
In Table 3, the “0\0” etc. notation refers to the (D plate \ E plate) drive setting in Table 2. The first column is for radiation mostly from the E plate. At 14.5 MHz, I ran out of attenuation to insert in the receiver line and the output voltage was very high. It peaked very sharply with frequency, at 14.454 MHz with 3 dB points only a few 10 of kHz wide! The minus numbers (shaded orange) represent the number of dB below the reference level the signal dropped to with all receiver line attenuation removed. I thought the CFA had finally begun acting like the inventors imply when they talk of “only a few degrees tolerance” in the drive phase. It was quite an exciting moment! I ran the signal generator level up and down: no jumps or artifacts that would suggest receiver oscillation. But then I started cross checking. I put in a foot of cable delay—no change. Then, tuned the receiver a few kHz. The peak moved a few (but less) kHz. I disconnected the D plate—no change!!! Next, I removed the CFA and connected the receiver line to the signal generator. The peak was still there at the same place. Pretty powerful CFA to work disconnected!!
I believe that the receiver antenna trimmer is not large enough to fully compensate for a mis-tuned RF stage, so that even though it was peaked at each frequency, the receiver doesn’t get to the same sensitivity at each frequency. As long as the frequency is constant, as it is within each table 1 or 2, the process works. The gain of the “detector” merely scales the number of dB attenuation required to get to the reference level. The receiver is now apart, connected to a sweeper, and getting a thorough retune.
No Sweet Spot So Far
So far I have not found any tuning “sweet spot” in any of these experiments. As always, the antenna keeps acting like the sum of two antenna elements adding radiation linearly. The large amount of interaction that seems to appear as the CFA is made bigger in terms of wavelength suggests another direction of thought. Could this strong interaction and large circulating current be “recycled” by a correct tuner to contribute again to the radiation?? Might that increase the efficiency? This has been suggested by other researchers, I believe. Is the CFA not a unique radiator but instead an ordinary radiator uniquely hung on a tuner? Perhaps Spice can shed some light on this as I continue in search of radiation!
Originally posted on the AntennaX Online Magazine by Joel C. Hungerford, KB1EGI
Last Updated : 29th May 2024