At age twenty-nine and already one of the most well-known radio engineers in the world, Edwin H. Armstrong was a veteran of the great war, and the president of the Radio Club of America. He was also professor of electrical engineering at Columbia University in New York City, where the R.C.A. was based and met regularly. Later recognized as one of the most important inventors in radio, Armstrong embodied the close relationship between amateur experimenters of the early years and leading academic and commercial radio researchers. In Armstrong’s case they were the same person.

His article “A New Method for the Reception of Weak Signals at Short Wavelength”¹ led the February 1920 issue of QST. It described a new receiving technique especially useful to hams of the time who were struggling with how to make use of signals that suffered two afflictions:  they were very short in wavelength and very weak. He had developed the method while in the Army during the war.

Having presented the material at one of the R.C.A.’s regular meetings the previous December, he gave the ARRL permission to reprint it under an agreement with the League that dated back before the war to regularly publish such talks.²

You get a strange feeling reading an article such as this. Looking in from the twenty-first century, you know how the technological story eventually unfolds in the years and decades that follow. You can’t help but empathize with one of the great minds of the time as he struggles to extend the boundaries of the radio art, on the verge of discovering a new solution, being held back by the inadequacies of the available equipment.

Armstrong, an experienced author of peer-reviewed technical publications, concisely described a challenge that was of great interest to amateurs who wondered about the usability of wavelengths below 200 meters, the only direction for expansion available to them. His objective was to build a receiver that covered a wide range of wavelengths (600 to 50 meters), was easily tunable, and faithfully reproduced the characteristics of the transmitted signal. This last requirement meant that the receiver must not distort the original modulation of an undamped signal—or if a spark (damped) signal, it should reproduce the original note, the personal sound of an individual transmitter.

He explained that there were three usual direct methods of reception, none of which were satisfactory: (1) rectification (detection) of the radio-frequency (RF) signal, as in a crystal set, followed by amplification of the resulting audio; (2) reordering things to first amplify the RF signal followed by rectification; and (3) heterodyning directly down to audio frequencies. Heterodyning means mixing the original radio signal with another signal produced within the receiver but having a slightly higher or lower frequency. This produces an audio signal with a frequency that is the difference of the two mixed signals.³ If the two frequencies are close enough, you’d get a frequency (called a “beat” frequency) that was low enough to be audible to the human ear.

Since rectification was inherently inefficient at low signal levels, the first approach worked poorly with weak signals—and weak signals were what you most wanted to hear since the signals coming from the farthest distance would also be the weakest. To make things worse, low-frequency (audio) amplifiers in 1920 were inherently noisy, limiting how much amplification you could employ, and how much useful signal you could get in the headphones.

In the second method, rectification was more efficient with an RF signal that had first been amplified. However, inter-electrode capacitance in vacuum tubes confined their operation to longer wavelengths in simple circuits. If you added inductance to counter this effect, you would end up with a receiver that was cumbersome to tune and prone to internal oscillations.

The third method, heterodyning to get audio, seemed to be the best alternative, but in 1920 it was impractical because of the instability of the beat note; vacuum tube oscillators were quite unstable at short wavelengths (high frequencies) in both receivers and transmitters.

Armstrong introduced an in-between approach:  heterodyne the original RF signal down to a frequency he described as “substantially above audibility,”⁴ in this case 100,000 cycles per second (or kc⁵), then pass it through amplification and rectification. The big advantage of this method was a greatly increased rectification or detection efficiency made possible by a much higher signal level after amplification. However, to use this for CW (undamped) reception, one must somehow modulate an unmodulated, undamped signal⁶ to make it audible since the detected frequency would be too high to hear.  Even though a stable oscillator could be built for this lower frequency, using one to further heterodyne this intermediate frequency down to an audible one would not work since the resulting note would still depend on the stability of the oscillator used in the previous heterodyne step.

Once a desired operating frequency was chosen and tuned (he used 3 Mc, or 3 million cycles per second) the heterodyne signal, or local oscillator in today’s terminology, could be adjusted for the most efficient rectification in the first stage, which, he wrote, was a non-critical adjustment. Of course it was only non-critical and did not need to be changed if one intended to stay on one frequency. Selectivity could be improved if the amplification stage was made regenerative and the coupling between stages was tuned.

As for spark (damped) signals, previous attempts using heterodyning resulted in the loss of the signal’s original note—the characteristic buzzy tone produced by the frequency with which sparks were produced. The signal’s note was lost because the efficiency of rectification (or mixing) in heterodyning depends strongly on the phase difference of the two signals being mixed. In spark, each “wave train” (each burst of RF energy after a spark) from the transmitter typically begins at a different, random phase angle, or point in the sinusoidal wave. Thus each wave train is heterodyned to a different, random level.  Also, since the audio frequency beat note is so low compared to the (RF) frequency in both the wave train and the local oscillator, there is no time to complete even part of one cycle of a beat note.  Both effects result in a random hissing sound at the receiver when listening to a spark signal, difficult to listen to and decode and difficult to distinguish from other nearby signals.

With a much higher beat note as in his new design, several beats per wave train could be produced, the phase difference would therefore vary through several cycles, and so the initial phase of each wave train mattered much less. The note of the original spark signal was thus preserved. In fact, to the listener, the difference in the characteristic notes of individual transmitters was even more noticeable than when using a simple crystal detector.

Armstrong goes on to describe other variations of this method of reception, including using multiple stages of different intermediate frequencies.  The basic idea of using an intermediate frequency would become standard practice and have many subsidiary benefits in receiver design, perhaps only some of which Armstrong himself understood fully at this early point. More than ninety years later these basic principles can be found in the design of a majority of receivers of all sorts, including cell phones and satellites, garage door openers and wireless computer networks (not to mention actual radios and televisions).  While acknowledging the work of others upon which he had built, he asserted that “the application of the principle to the reception of short waves is, I believe, new and it is for this reason that this paper is presented.” That is, presented to the people who needed it most and prompted its development: his fellow amateurs.

Though constrained by crude components, Armstrong was changing the fundamental nature of the radio art before his thirtieth birthday.

de W2PA