16-02-2017, 11:07 PM
As I understand it, the pilot tone, which was allowed 10% of the available deviation, was placed at 19 kHz to aid in its noise-free recovery. Given the triangular noise spectrum of FM, then for a given deviation (±7.5 kHz in this case), the noise at 19 kHz was 6 dB lower than the noise at 38 kHz. Also, at 19 kHz, the pilot tone is surrounded by a reasonable amount of “white space”, so that the circuits used to select, amplify and process it were unlikely to be affected by nearby signals. On the other hand, at 38 kHz, it would have been surrounded by nearby difference signal sidebands that might well have been significantly greater in magnitude, and might therefore upset recovery circuits. Conceivably a pilot tone at 38 kHz would have required greater amplitude, and something like 20% of the available deviation to be adequately robust, thus leaving 80% rather than 90% of the deviation available for the programme signals.
Back in the late 1950s, I think that the assumption would have been made that receivers would most likely have used sharply tuned LC filters to extract the pilot tone. So keeping the pilot tone clear of noise and nearby sidebands was a key desideratum. PLL techniques, which might be better able to deal with adverse situations, were known and doable with valves (and to some extent were used in colour TV receivers for colour subcarrier recovery), but were seen as too complex and too costly for general use in domestic receivers. In fact PLL-based decoding did not arrive for general use in FM stereo receivers until the IC age, starting with the RCA CA3090 in 1971 and soon followed by the Fairchild µA758 and Motorola MC1310. And these were second-generation IC decoders, the first generation having been represented by the Motorola MC1304/5.
Note though that the American MTS system also included a pilot tone, even though by then it could assumed that PLL-type decoders would be used. In this case the pilot tone was at line frequency, namely 15 734 Hz. So clearly the view was that despite circuit improvements over 20 years or so, a half-frequency pilot tone was still preferable.
Cheers,
Steve
Back in the late 1950s, I think that the assumption would have been made that receivers would most likely have used sharply tuned LC filters to extract the pilot tone. So keeping the pilot tone clear of noise and nearby sidebands was a key desideratum. PLL techniques, which might be better able to deal with adverse situations, were known and doable with valves (and to some extent were used in colour TV receivers for colour subcarrier recovery), but were seen as too complex and too costly for general use in domestic receivers. In fact PLL-based decoding did not arrive for general use in FM stereo receivers until the IC age, starting with the RCA CA3090 in 1971 and soon followed by the Fairchild µA758 and Motorola MC1310. And these were second-generation IC decoders, the first generation having been represented by the Motorola MC1304/5.
Note though that the American MTS system also included a pilot tone, even though by then it could assumed that PLL-type decoders would be used. In this case the pilot tone was at line frequency, namely 15 734 Hz. So clearly the view was that despite circuit improvements over 20 years or so, a half-frequency pilot tone was still preferable.
Cheers,
Steve
