Rich,
Thank you for asking this particular question, because I anticipated that this subject may cause confusion. Please note that in the statement you quoted, I used the word "can" instead of "will." This is because, depending on the tuning circuit, a short antenna can present either a low or a high impedance. I will use the dummy antenna I use frequently, 30 pF in series with 46 ohms, as an example. 30 pF represents a typical capacitance of a three-meter vertical antenna. 46 ohms represents the rather poor antenna resistance I expect a typical Part 15 AM installation to have.
(Again, I said that this is my guess of the ground resistance, since I have no way to measure it. This is not because of a lack of measurement equipment. I have an ancient GR 1606B RF bridge, which, in principle, can be used to make the measurement, but there is simply no place to put the instrument (or myself). If the bridge had small dimensions, and could be remotely operated and read, the measurement could be easily made. However, with my equipment, the operator has to adjust for a null to make a measurement, and both the bulky instrument and the body of the operator would interfere with the measurements. I have read an article about a group of hams making vertical antenna input impedance measurements by constructing a counterpoise over a hole in the ground. They got into the hole with their instruments to make their measurements. This kind of procedure is not applicable to Part 15 AM, however. Since you are a broadcast engineer with many years of experience, perhaps you might have suggestions about how to make the measurement.)
By using a loading coil in series with the antenna, the 30 pF antenna capacitance is tuned out, leaving the ground resistance of 46 ohms, plus the RF resistance of the loading coil. However, there are transmitter designs, such as the Wenzel and the Rangemaster, that do not use a loading coil located at the base of the antenna. (I remember one of your posts where you referred to this fact yourself.) In these designs, the loading coil is not simply moved from the antenna to the transmitter box. There is a fundamental difference in the way the antenna is tuned. In Wenzel/Rangemaster tuning, the secondary winding of a step-up RF transformer forms a parallel tuned circuit with the 30 pF antenna capacitance. The inductance can be set with taps on the secondary winding, and fine tuning is done with a trimmer capacitor in parallel with the antenna input. In this case, the antenna resistance is high. At 1.5 MHz, the 46 ohm series resistance appears as approximately 270 k ohm parallel resistance. The parallel equivalent resistance is roughly the capacitive reactance of the antenna multiplied by the antenna Q.
A Part 15 tube transmitter output circuit needs an inductor at its output to resonate with the 30 pF antenna capacitance. However, just like in the Wenzel and Rangemaster designs, the inductor forms a parallel resonant circuit with the antenna capacitance, (there is a trimmer capacitor across the antenna input terminals) and the resistive load impedance seen by the antenna is the parallel equivalent resistance, which is high.
First, I'll give some background about what I have been doing in relation to this subject:
It was around 1958 that I first built a Part 15 transmitter. It was a one-tube "phono oscillator" that used a 1.5 VDC cell for the "A" battery and a 22 1/2 VDC "B" battery. The tube was either a 1T4 or a 1L4; I don't remenber which. The antenna was a ten-foot length of wire. There was no ground at all. A phono cartridge supplied enough voltage to modulate the grid. At that time, I thought that this is why it was called a "phono oscillator." Recently, I found out from a post by Neil that this name was given to this sort of device more than twenty years earler (in the 1930s) because it was used to allow a phonograph to be played through the audio amplifier of an AM radio receiver. To transmit voice, an inexpensive crystal microphone did not supply nearly enough voltage to modulate the grid. It was necessary for me to pay what was a lot of money for me at the time for a carbon microphone and a microphone transformer to get my phono oscillator to work with voice. I was able to transmit to an AM radio in my house.
This was actually pretty exciting for a while, but then it finally got old, and I went on to do other things.
It was more than forty years later that I became interested in Part 15 again.
In the April 1999 issue of RF Design, Tom Warnagaris of Southwest Research Institute published a comprehensive article about Part 15 LF, MF, and FM. This is the only professional-level article I have ever seen about these three Part 15 bands, which are often used by hobbyists. The sort of Part 15 articles I have seen in trade magazines have usually been related to the UHF and microwave bands that are of interest to big business. Most engineers, after all, work for big business. What attracted my attention was that Warnagaris supplied graphs describing what he called expected range in these hobby bands. In the 160 to 190 kHz band, he predicted a maximum range of 2.12 miles. He predicted 5.48 miles at 510 kHz, 49.2 miles at 1700 kHz, and 1668 feet at 98 MHz. Obviously, these predicted ranges are outrageously long, but the calculations were correct. The assumptions that were made, however, were overly optimistic. He assumed a perfectly conducting flat earth (no groundwave losses, and no over-the-horizon propagation problems), and a receiver limited only by rural daytime atmospheric noise (a big antenna, and no urban noise). The high loss in a 3-meter transmitting antenna was taken into account. He assumed an antenna Q of 250. For FM, he assumed the thermal noise limit at the input of the receiver. He also assumed that the transmitter efficiency was 75%, which he considered to be easily achievable. Clearly, he had little actual experience with Part 15 AM, because such an efficiency, while easy to achieve with some transmitters, is actually very difficult to obtain with Part 15 AM transmitters.
It was obvious to me that the ranges were overestimates. A 49.2 mile range is difficult to obtain, even for a full-power AM transmitter. But, even so, this publication got me interested in Part 15 AM again.
It was another two years before I actually started doing something about Part 15 AM. I had clippped the Warnagaris article and it was stored in my files. In April of 2001, I finally started to construct a Part 15 transmitter. Looking through my old issues of Radio-Electronics, I found an article by Richard A. Nelson about a 1-watt 160-190 kHz Part 15 transmitter in the September 1989 issue. It was a simple matter to adapt the design for 100 mW part 15 AM. Nelson's design used an IRF 511 MOSFET driven at the gate by logic. The drain drove the primary winding of an RF transformer with a high turns ratio. The output winding had a tuning capacitor to ground for tuning the antenna. I later found out that that this is also the basic architecture of the Rangemaster transmitter. This is why I now call it the Nelson/Hamilton transmitter. I am not suggesting that Hamilton got the circuit from Nelson. The circuit is simple, and any number of people could have thought of it, probably even before 1989.
It did not take me long to realize that the MOSFET in Nelson's circuit was going to be a problem. The parasitic diode from drain to source was absorbing energy and reducing efficiency considerably. I altered the circuit so that I was able to use a JFET, which, of course, is a depletion-mode device, and cannot be driven directly by logic. The JFET gave me better efficiency than the MOSFET. After a few months of working with Part 15 AM both practically and theoretically, I concluded that, not only can't you get several miles of range, as Warnagaris suggested, but you can't even get one mile of reliable 24/7 range, even if you somehow managed to convert the full 100 mW of input power to the final stage into radiated RF power. Also, the inefficiency of the 3-meter antenna is so great, that you will not be able to convert even one mW of the 100 mW input power into radiated power. So, once again, my Part 15 efforts went back into a deep sleep.
Eventually, I returned to Part 15 again by simply lowering my expectations. I realized that the FCC was not going to allow an unlicensed service that has significant range. LPB advertised only a 1/4 mile range for their AM-2000. This was the best Part 15 AM transmitter when it was still being manufactured. It was pricey, too--about twice as much as a Rangemaster today. To get even this modest range, LPB recommended "elevating" the antenna--to get the signal over obstructing structures, of course. I think that it might be possible to get a significant radiated power improvement over existing Part 15 AM systems with 3-meter antennas without doing any elevating. This could be done by improving transmitter efficiency and making slight improvements to the antenna that increase efficiency by a few dB. This could make a 3-meter installation the equivalent of a 30-foot installation, or even higher.
Really good technical information is difficult to find for Part 15 AM. The Warnagaris article I referred to is the best I know about, and it isn't very good. For that reason, I want to provide some practical information relating to circuit and system design that is not available elsewhere.
The principal difficulty with the Nelson/Hamilton design is that the tuning network does not provide good impedance matching of the final stage of the transmitter to the antenna. The SSTRAN is actually better with this respect. In this thread, I have been using 46 ohms in series with 30 pF as a dummy antenna for testing. This dummy load represents a typical 3-meter antenna with a 46 om ground resistance. If the ground resistance were improved to 23 ohms, the antenna efficiency would double, but the efficiency of the the Nelson/Hamilton design would be halved. This would result in no improvement to the radiated power from lowering the ground resistance. The SSTRAN, however, would be capable of getting nearly the same efficiency with either the 23 ohm resistance and the 46 ohm resistance. So, someone with a really good ground system should use the SSTRAN (despite the fact that it is difficult to tune) instead of the Rangemaser. The tuning difficulty with the SSTRAN relates to the fact that its antenna tuning inductor has the double function of resonating with the antenna capacitance, and providing the inductance of an impedance-matching "L" network. Despite this awkward arrangement, the SSTRAN has an additional degree of freedom in impedance matching that the Rangemaster does not have, giving it the advantage of being able to tune to a greater variety ground resistances. Both the Rangemaster and the SSTRAN have roughly the same efficiency with 46 ohm ground resistance, but the SSTRAN would give more radiated power with smaller ground resistances.
The original Nelson circuit uses a tightly-coupled air-core RF transformer. An air-core transformer has the advantage that its inductance does not change as the MOSFET drain current changes. The Rangemaster uses an iron powder toroid to save space, but this has the disadvantage that the RF transformer is difficult to tune if it is far out of adjustment. When the final stage is badly out of tune, the MOSFET DC drain current is high, causing the inductance of the transformer to be low. Once the transformer is nearly correctly tuned, howevever, final adjustment is quite easy. I found that, when the three ZVNL110A MOSFETs of the Rangemeaster are in the circuit, tuning the output cap for minimum drain current also tunes for highest efficiency, which is about 25 % when the dummy antenna has a 30 pF capacitor and a 46 ohm resistor in series. When the three MOSFETs in parallel are replaced by an IRF 511 (or, equivalently, a IRF 510), as was used in the Nelson circut, a 25 % efficiency is also obtained, but tuning is not so straighforward. I had to look at the drain voltage with an oscilloscope to get the waveform to look good. In particular it was necessary for the waveform during the "on" time of the cycle to be horizontal and near zero volts. Changing the resistor of the dummy antenna to 23 ohms reduces the efficiency by about a half. When a real antenna is used, the antenna efficiency increases when the ground resistance is decreased, but this improvement in antenna efficiency resulting from a good ground does not help a Rangemaster user. The radiated power remains the same whether a good ground or a bad ground is used.
I was able to more than double the 25 % efficiency when I was using an IRF 510 by connecting a a large capacitor (more than 1000 pF) from drain to ground. This caused the multiple pulses seen at the drain during each cycle to become one pulse, and the parasitic diode from drain to source was not forward biased during the entire cycle. When I tried this with the three ZVNL110As in parallel, the trick did not work. Although there was again only one pulse, the parasitic diodes of the three MOSFETs were forward biased each cycle, anyway. I had to use more drive to power the gate of the IRF 510 than the three gates of the ZVNL110As, because the IRF 510 has a higher threshold voltage. For the IRF 510 gate, I used a 7408 AND gate with an 820 ohm pullup resistor to the +5 VDC supply. For the three ZVNL110As, I used a 74HCT08 AND gate with a 47 ohm resistor in series with the three MOSFET gates in parallel that it was driving.
I checked the effect of the 1N5819 Scottky rectifier diode used as a snubber across the parasitic drain to source diodes of the MOSFETs. I found that the snubber helped improve efficiency, from about 19% to 25%. The original Nelson circuit had no snubber.