In a typical TV chassis, a single reference QSS mixer is used to synchronously demodulate the IF to give the sound and NICAM carriers. It uses a quadrature reference (shifted by 90 degrees), generated from the video demodulator. When the IF signal is multiplied by this reference, the video information is nulled leaving so that only the sound and NICAM carriers are demodulated.
Are there any papers on the net which explain this in more detail? The part I can't understand is if the sound carrier (FM) is added to the baseband video, before modulating a UHF carrier, why shouldn't this be nulled out as well! What am I missing?