I have since found a 1976 Motorola (Quasar) paper (1) that addresses the issue of distortion in the vision demodulation process attributable to phase modulation of the reference carrier caused by the Nyquist slope. Avoiding this problem would require either a very narrow bandwidth reference channel (just a small number of fh sidebands wide) or re-establishing double sideband integrity before limiting. The first option could be realized with PLL fully synchronous demodulation. At the time the paper was written it was thought that use of a suitable SAW filter in the reference channel would allow the second option.
Evidently though eliminating the Nyquist slope error in quasi-synchronous vision demodulator ICs was not a major concern until the arrival of multichannel sound. For example, it is not really mentioned as an issue in the papers dealing with the MC1330, the TBA440 (2). With the later MC1331 (4), Motorola seemed to be concentrating on detail improvements to the multiplier operation, along with the use of a separate intercarrier multiplier as a way of reducing the colour subcarrier-sound carrier beat. One might reasonably infer that the benefits obtainable with the standard designs, as compared with diode demodulators, were viewed as sufficient for most domestic receiver purposes, and anyway there was always the option of using fully-synchronous demodulation.
The arrival of multichannel sound forced the issue, and then for the sound demodulation process only, in order to avoid intercarrier buzz. This is discussed in detail in a 1981 Zenith paper (5). Suggested solutions were quasi-split sound (which by 1981 might already have been in use in Europe and/or Japan) and split sound using a narrow sound IF filter followed by a 2nd independent conversion to 4.5 MHz. (In practice a 10.7 rather than 4.5 MHz 2nd sound IF seems to have been used by Sony and others where this scheme was adopted.) But the split sound system was vulnerable to incidental phase modulation (IPM) on the signal as received so receivers using it typically had a conventional intercarrier sound channel for use in such situations.
Generally it seems that quasi-split sound ICs were used alongside conventional vision IF and quasi-synchronous demodulator ICs, inclusive of their Nyquist errors. I am not aware of any vision quasi-synchronous demodulator ICs that had a separate pathway from a SAW filter carrier output for the vision reference channel.
With professional equipment it may have been different, and for example the BBC RC1/511 receiver mentioned in post #3 “reconditioned” the reference channel to eliminate the Nyquist slope. The RC5M/503 UHF rebroadcast receiver (information available at
http://www.bbceng.info/Designs/desig...EDI_sheets.htm) seems to have been similarly equipped, judging by the commentary:
“Selectivity and Nyquist shaping are obtained by the use of a specially developed surface acoustic wave (SAW) filter, and no IF alignment is required. A second SAW filter extracts the vision IF carrier, and amplitude modulation is removed by low-phase shift limiters; the resulting carrier is used to demodulate the vision signal. This “exalted-carrier”, or “pseudo-synchronous”, demodulation means that the effect of any incidental phase modulation (IPM) present on the input is greatly reduced.”
The RC5M/503 was intended to replace the RC5M/501 and RC5M/502, and the latter apparently had fully synchronous demodulation, it being stated:
“Synchronous detection is used, thereby removing quadrature distortion, giving improved pulse response, and a considerable reduction of chrominance-luminance crosstalk. Differential gain and phase distortions are also much reduced”.
My take is that the RC5M/502 used a PLL fully synchronous demodulator, and that it was being compared with the preceding RC5M/501, for which synchronous demodulation is not mentioned, so one assumes that it had envelope demodulation.
The LM1823 application note mentions the need for a wider (500 Hz instead of 73 Hz) PLL bandwidth when the incoming vision signal suffers from incidental phase modulation. But 500 Hz maybe enough for Nyquist slope errors to be non-negligible. So it seems that the approach used in the BBC RC5M/503 accommodates incidental phase modulation whilst avoiding the Nyquist error.
(1) Rzeszewski, “A System Approach to Synchronous Detection”, IEEE Transactions, 1976
(2) Schatter, “Monolithic TV IF System TBA440”, IEEE Transactions, 1972
(3) Wilcox, “A new TV Video/Sound Detector IC”, IEEE Transactions, 1974
(4) Fockens & Eiler, “Intercarrier Buzz Phenomena Analysis and Cures”, IEEE Transactions, 1981
Cheers,