[old_tv_nut]

That Dimensia IQ demod design was ingenious in that they found a way to make an I channel delay line with only one IC pin (plus one external transistor). As an analog IC designer, I thought that was one of the neatest tricks I ever saw. Essentially, they tied the input of the delay line to ground, then drove the delay line output and what would normally be the delay line ground up and down with the input "I" baseband voltage. They then used the same driving pin to sense the output current!

Despite my admiration for the IC design, When I tested one of those sets for chroma resolution (working at Zenith), it was essentially no better than the R-Y/B-Y sets. The I channel high frequencies were quite attenuated. It appeared that there was too much phase distortion in the IF to allow the I channel high frequencies to be boosted to a useful level. I went through some effort to design a better IF for a possible wideband design myself, but concluded that the NTSC got away with their system design partly because tests were done on small screens. My basic conclusion was that NTSC needed a more symmetrical Q filter in the transmitter - it should have traps on the upper and lower edge that match the 4.5 MHz sound trap (+/- 900 kHz from the chroma carrier). If the Q sidebands transmitted would be sharply curtailed beyond +/- 500 kHz, there could be no quadrature crosstalk from Q into I, and it would behave like the NTSC intended - but NTSC Q filter specs were too broad, and any reasonable IF would show quadrature distortion and therefore the I response could not be boosted to the theoretical value. I proved this by modifying the filters in an NTSC encoder -and when we saw how well it worked, we gave up trying to design an I/Q set, since there was no way to go back and modify every encoder out there.