by JOHN K. MITCHELL
NOT A MINOR MAJOR
In the July 1986 "Audio ETC" column, I reminisced about the introduction of high-fidelity audio to the general public by the A.T.&T. Co. about 50 years ago. I suspected that it was to create a demand for broad-band (50 Hz to 8 kHz) radio lines, which would have overstepped the legal limit of 5 kHz. The effort was successful, and a short time later, with the advent of frequency modulation (FM), even higher quality lines were in demand. But what were the roots of FM? Major Edwin H. Armstrong was a prolific inventor, and his innovations in the art of radio transmission and reception were numerous. (If I remember rightly, it was he who invented the superheterodyne method of reception.) In 1936, Armstrong proposed the use of frequency modulation to reduce disturbances in radio signals. His paper, titled "A Method of Reducing Disturbances in Radio Signaling by a System of Frequency Modulation," was published in the May 1936 Proceedings of the I.R.E. I attended a meeting where the subject was presented to the members. I don't know whether Armstrong had experimented with the process before he published his paper, but when I first became acquainted with him a couple of years later, in connection with my duties for the telephone company, he was deeply involved in its development.
Until frequency modulation was authorized, radio broadcasts were by amplitude modulation (AM). The highest frequency that could be reproduced from these broadcasts was limited by law to 5 kHz. (It is now 15 kHz per channel, and interference from the "spillover" is controlled by several factors related to signal strengths at the fringes of the coverage areas of adjacent-channel stations.) AM has many disadvantages, particularly the proclivity of an AM receiver to reproduce the audio portion of electrical noise in its vicinity. This can sometimes be so loud as to destroy the audio produced from the radio signal.
Another disadvantage (because of the particular radio frequency allocation) is the extensiveness of the ground wave, which, when mixed with the sky wave, will produce periodic fading of the signal when the sky wave is reflected by the ionosphere and arrives at a strength comparable to that of the ground wave. This is caused by the sky wave path being longer than that for the ground wave. Likewise, because sky waves travel farther at night, some broadcasters are required to reduce the strength of their signal at sundown, thus reducing the size of their audience.
Major Armstrong had been searching for a solution to these problems and had been authorized to operate W2XMN, an experimental station at Alpine, N.J., in conjunction with another station at Paxton, Mass. Both were on high elevations, and their locations were ideal for the type of propagation measurements being made. They were a little more than 100 miles apart and just within line of sight of each other, which was a necessary requirement for the radio frequencies that were involved.
At the time Armstrong was experimenting, very little was known about the propagation characteristics of VHF signals (30 to 300 MHz). Various experiments in the region of 100 MHz had shown the feasibility of using such frequencies for aircraft guidance during landings, but not much progress had been made in examining the long-distance capabilities for communicating between ground stations. Likewise, there was very little in the way of transmitting tubes for frequencies much above 60 MHz. The only one I can remember was the G.E. ZP-2, which would produce about 100 watts at 90 to 100 MHz. I believe it was used by the Bureau of Standards at College Park, Md. in the early '30s for experiments in producing glide-slope guidance signals at 93.7 MHz.
In 1936, when Armstrong was conducting his experiments, the F.C.C. allocated frequencies only as high as 60 MHz, and even at that there were no allocations between 30 and 56 MHz.
The 56-to-60-MHz band was set aside for experimentation and for use by amateurs. It must have been that band in which Armstrong's experiments were conducted. Although I had many conversations with Paul DeMars, the W2XMN station manager, during my frequent visits to try to improve the noise figure of the 15-kHz lines, I cannot remember any specific frequency being mentioned. I do remember that Armstrong pleaded for assignments above 60 MHz and was very disappointed when the F.C.C. assigned the band 42 to 50 MHz for FM. (There were five assignments from 42.1 to 42.9 MHz for educational purposes and 35 channels from 43.1 to 49.9 MHz for commercial use. Each channel was 200 kHz wide.) In Armstrong's experimental transmissions between Alpine and Paxton, he had undoubtedly discovered that the Kennelly-Heaviside layer (which we now call the ionosphere) had little effect at the frequencies he was using.
Furthermore, I think he had experimented with enough frequencies to realize that the higher he went the less effect was obtained from ionospheric reflections. In fact, we now know that frequencies above 60 MHz penetrate the ionosphere and are not subject to "skip." It was unfortunate that FM had to suffer reflections from the sporadic E layer for the next decade.
It wasn't until the results were in from the experience of World War II that justification for Armstrong's contentions were obtained. According to Paul DeMars and others I have talked to, Ed Armstrong had a running debate going with Kenneth Norton, who was then chief engineer of the F.C.C. but who later moved to the Bureau of Standards and became known as an authority on radio wave propagation through the many papers he wrote on the subject.
It was Armstrong's contention that FM transmitters located 120 miles apart could operate on the same carrier frequency without interfering with one another. Even a fringe-area listener who was able to hear either signal would hear only the stronger signal because of the capture-effect peculiarity of the detector.
In 1947, the International Telecommunications Conference held in Atlantic City, N.J. resulted in the world being divided into three distinct areas for the allocation of radio frequencies from 10 kHz to 10,500 MHz. That convention reassigned the frequency-modulation broadcast band, making it 88 to 108 MHz instead of 42 to 50 MHz, and increased the number of channels to 100. The prime result of the reallocation was to improve the long-distance reception quality of FM, but it introduced a different difficulty that we now call multipath, produced by buildings and other ground obstructions. One of the nice things that accrued from the higher frequencies of 88 to 108 MHz was the folded dipole that could be tacked inside tuner cabinets by manufacturers. This made it unnecessary to acquire a special aerial, which could have been a deterrent to the sales of AM/FM/phono consoles at that time.
In the early days of FM, some hobbyists discovered that a low-Q resonant tank circuit could be used as a slope detector to demodulate the signal. However, these were not very satisfactory, and other, more sophisticated methods were needed. The first FM tuner by G.E. used a Foster-Seeley discriminator, consisting of two diodes operated in push-pull. Limiters ahead of the detector controlled the variations in signal amplitude, and the tuner gave excellent results. I obtained one of these before they were released for sale by joining a group of WOR engineers in a testing program. Charles Singer, the station manager at WOR's transmitter, invited me to join the group because I was working on one of their radio lines at the time. If I remember rightly, that tuner cost me $27. G.E. later furnished a new discriminator transformer and a couple of condensers to overcome a minor fault.
The tuner was quite bulky compared to things of today. It looked like one of the early table radios, but it had no speaker. To the best of my memory, its dimensions were 6 inches high by 6 inches wide by 12 inches deep, and it was housed in a nice walnut (or was it rosewood?) cabinet. Its tuning dial was straight line and its antenna was provided by the line cord. It had one stage of amplification and was connected to the AM receiver at the Aux jack. The mode switch on the receiver controlled its output, but a separate switch turned it on.
R.C.A. developed a discriminator that was not affected by carrier amplitude variations. It was called a "Ratio Detector," and it too used diodes in push-pull, but it responded only to the changes in the ratio of the voltages across the two diodes, whereas the output of the Foster-Seeley discriminator in the G.E. tuner consisted of the difference in the voltages across them.
Mention of R.C.A. brings to mind the court battles that consumed most of Major Armstrong's later life. Many manufacturers infringed on his patents, and he was in court frequently. In all cases except one he was vindicated, even besting Western Electric. The one case that dragged on and on was that which involved R.C.A., although I understand that was finally settled with his estate, out of court.
(Source: Audio magazine, Oct 1987)
= = = =