Looking back to the original question, G8HQP Dave’s comments have prompted a close re-read of the Lunn/Motorola article plus a search for other references.
In the Lunn article, after presenting the basic layout of the quasi-synchronous demodulator, as previously proposed by both Biolotti and Macario, the comment is made that the characteristics of the switching signal can be changed by selective circuits. This established the fact that selective circuits may be used, but are not an inherent feature of the scheme.
This is reinforced by the application note AN189 for the MC1496 multipurpose double-balanced multiplier, available at:
http://www.spelektroniikka.fi/kuvat/mc1496appl.pdf. Therein the basic AM demodulator case is presented as using limited, but not filtered signal carrier.
In the Lunn article, later on the basic premise is repeated as “..the signal and reference switching channels are separate so that the frequency response of the switching path may be operated on without affecting the main channel.” A key objective was to minimize the chroma-sound beat, and the original proposal included not only a tuned circuit peaked at vision carrier frequency but a notch at the chroma sub-carrier frequency. There was concern that too high of a Q in the tuned circuit might make receiver tuning difficult (a Q of 30 is mentioned) hence the anticipated need for the notch. In practice, for the MC1330 (and all that followed) a simple tank circuit seems to have sufficed. The datasheet recommends 20 to 50, with the comment that the higher the Q, the better the chroma-sound beat rejection.
Anyway, that indicates that the for the vision demodulator case, tank circuit is not there out of basic necessity, but primarily to reduce the chroma-sound beat.
Nothing is said in the article about the errors and adverse effects introduced by the Nyquist slope, but then that did not become an issue until higher quality sound was required and stereo sound was introduced, which saw the development of the quasi-split sound technique and in one or two cases reintroduction of split sound IF systems.
Some interesting comments may found in application note AN391 for the National LM1823 vision IF IC with PLL fully synchronous demodulation, available at:
http://www.national.com/an/AN/AN-391.pdf. Clearly this is “selling” the idea that the PLL technique is better than quasi-synchronous, but it does reinforce G8HQP Dave’s comment that unless tank circuit tuning corresponds exactly to carrier frequency, there will be phase errors that vary with the original amplitude modulation fed to the limiter stage.
Getting back to the Lunn/Motorola concept of operating on the carrier channel frequency response, it would appear that if one wanted to correct for the Nyquist slope and so reduce sound buzz on conventional intercarrier systems, this could not be done by slightly detuning the post-limiter tank circuit, but would need to be done separately ahead of the limiter.
Cheers,