Armstrong had realized that higher frequency equipment would allow them to detect enemy shipping much more effectively, but at the time no practical "short wave" (defined then as any frequency above 500 kHz) amplifier existed.
It had been noticed some time before that if a regenerative receiver was allowed to go into oscillation, other receivers nearby would suddenly start picking up stations on frequencies different from those they were actually transmitted on. Armstrong (and others) soon realized that this was caused by a "supersonic" heterodyne (or beat, as in acoustic beating) between the station's carrier frequency and the oscillator frequency. Mixing two frequencies creates two new frequencies, one at the sum of the two frequencies mixed, and the other at their difference. Thus, for example, if a station were transmitting on 300 kHz and the oscillator were set to 400 kHz, the station would be heard not only at the original 300 kHz, but also at 100 kHz and 700 kHz. This process is known as heterodyning.
Armstrong realized that this was a potential solution to the "short wave" amplification problem. To monitor a frequency of 1500 kHz, he could set up an oscillator to, say, 1560 kHz, which would down-convert the signal to a 60 kHz intermediate frequency, which was far more amenable to high gain amplification using triodes.
Superheterodyne circuits originally used the self-resonance of iron-cored interstage coupling transformers to filter the intermediate frequency. (Even today, Intermediate Frequency tuned circuits are referred to as IF "transformers" although they would more correctly be termed "coils" like the aerial and oscillator tuned circuits). In modern receivers electromechanical filters such as Ceramic resonators, Surface Acoustic Wave (SAW) or crystal-lattice filters are more likley used provide selectivity at the intermediate frequency. Early superhets used IFs as low as 20 kHz, which made them extremely susceptible to image frequency interference, but at the time the main interest was sensitivity rather than selectivity.
Armstrong was able to put his ideas into practice quite quickly, and the technique was rapidly adopted by the military; however, it was less popular when radio broadcasting began in the 1920s, due both to the need for an extra tube for the oscillator, and to the amount of technical knowledge required to operate it. For domestic radios, an alternative approach to Short Wave "Tuned RF" ("TRF") amplification called the Neutrodyne became more popular for reasons of simplicity and economy. Armstrong sold his superheterodyne patent to Westinghouse, who sold it to RCA, who monopolized the market for superheterodyne receivers until 1930.
However, by the 1930s, improvements in vacuum tube technology rapidly eroded the TRF receiver's advantages. First, the development of practical indirectly heated cathodes allowed the mixer and oscillator functions to be combined in a single Pentode tube, in the so-called Autodyne mixer. This was rapidly followed by the introduction of low-cost multi-element tubes specifically designed for superheterodyne operation, and by the mid-30s the TRF technique was rendered obsolete. Virtually all radio receivers, including the receiver sections of television sets, now use the superheterodyne principle.
Reception starts with an antenna signal, optionally amplified, including the frequency the user wishes to tune, fd. The local oscillator is tuned to produce a frequency close to fd, fLO. The received signal is mixed with the local oscillator's, producing four frequencies in the output; the original signal, the original fLO, and the two new frequencies fd+fLO and fd-fLO. The output signal also generally contains a number of undesirable mixtures as well.
The amplifier portion of the system is tuned to be highly selective at a single frequency, fIF. By changing fLO, the resulting fd-fID (or fd+fID) signal can be tuned to the amplifier's fIF. In typical amplitude modulation ("AM radio" in the U.S., or MW) home receivers, that frequency is usually 455 kHz; for FM receivers, it is usually 10.7 MHz; for television, 45 MHz. Although the other signals from the mixed output of the heterodyne are still present when they reach the amplifier, they are either filtered out or simply left un-amplified.
The advantage to this method is that most of the radio's signal path has to be sensitive to only a narrow range of frequencies. Only the front end (the part before the frequency converter stage) needs to be sensitive to a wide frequency range. For example, the front end might need to be sensitive to 1–30 MHz, while the rest of the radio might need to be sensitive only to 455 kHz, a typical IF. Only one or two tuned stages need to be adjusted to track over the tuning range of the receiver; all the intermediate-frequency stages operate at a fixed frequency which need not be adjusted.
Sometimes, to overcome obstacles such as image response, more than one IF is used. In such a case, the front end might be sensitive to 1–30 MHz, the first half of the radio to 5 MHz, and the last half to 50 kHz. Two frequency converters would be used, and the radio would be a "Double Conversion Super Heterodyne"—a common example is a television receiver where the audio information is obtained from a second stage of intermediate frequency conversion. Occasionally special-purpose receivers will use an intermediate frequency much higher than the signal, in order to obtain very high image rejection.
Superheterodyne receivers have superior characteristics to simpler receiver types in frequency stability and selectivity. They offer much better stability than Tuned radio frequency receivers (TRF) because a tuneable oscillator is more easily stabilized than a tuneable filter, especially with modern frequency synthesizer technology. IF filters can give much narrower passbands at the same Q factor than an equivalent RF filter. A fixed IF also allows the use of a crystal filter in very critical designs such as radiotelephone receivers, in which exceptionally high selectivity is necessary. Regenerative and super-regenerative receivers offer better sensitivity than a TRF receiver, but suffer from stability and selectivity problems.
In the case of modern television receivers, no other technique was able to produce the precise bandpass characteristic needed for vestigial sideband reception, first used with the original NTSC system introduced in 1941. This originally involved a complex collection of tuneable inductors which needed careful adjustment, but since the early 1980s these have been replaced with precision electromechanical surface acoustic wave (SAW) filters. Fabricated by precision laser milling techniques, SAW filters are much cheaper to produce, can be made to extremely close tolerances, and are extremely stable in operation.
The next evolution of superheterodyne receiver design is the software defined radio architecture, where the IF processing after the initial IF filter is implemented in software. This technique is already in use in the latest design analog television receivers and digital set top boxes, where there are no coils or other resonant circuits used at all. The antenna simply connects via a small capacitor to a pin on an integrated circuit and all the signal processing is carried out digitally. Similar techniques are used in the tiny FM radios incorporated into Mobile phones and MP3 players.
Radio transmitters may also use a mixer stage to produce an output frequency, working more or less as the reverse of a superheterodyne receiver.
Drawbacks to the superheterodyne receiver include interference from signal frequencies close to the Intermediate Frequency. To prevent this, IF frequencies are generally controlled by regulatory authorities, and this is the reason most receivers use common IFs. Examples are 455 kHz for AM radio, 10.7 MHz for FM, and 45 MHz for television. Additionally, in urban environments with many strong signals, the signals from multiple transmitters may combine in the mixer stage to interfere with the desired signal.