Quote:
Originally Posted by Tomcomm
Did RCA or any other tubed roundy maker return to the original NTSC wideband, low level chroma demodulators?
|
Tube demodulators actually came in three flavors in terms of bandwidth. The original I/Q demods in the CT-100 were of course low level and of greatest possible bandwidth.
Demodulators that went directly to the CRT grids (like the (R-Y)/(G-Y) in the CTC-5 "Super chassis) were probably the narrowest because of needing high load resistance to get the necessary gain in a single stage. The X-Z were somewhere in between. They could not have too much bandwidth, or they would show bad color edges due to I/Q quadrature crosstalk; but they would not be so limited by the capacitive rolloff, due to having some gain in the demods and some in the matrix amplifiers.
I am quite familiar with the 80's attempt at I/Q demodulation in the RCA solid state set, as I studied it closely while looking at the possibility for widebanding at Zenith. The fact is, creating an IF that passes the chroma without introducing phase distortion of the upper sideband can be quite difficult (although surface-wave IF filters could probably be designed to do it). Unfortunately, RCA did not succeed, and the high-frequency I channel gain on that set was so low in order to hide problems that it did little to improve the chroma resolution. It could fairly be said to be present in the set for advertising purposes, but that was about it. It's too bad, because some very clever design went into the analog color chip to minimize pin count while sitll providing the required I-channel delay line. It is one of my favorites. The designers came up with an amazingly clever way to drive and sense the delay line with a single pin! However, when we put the set side by side with an equiband Zenith (which had a very simple color takeoff with a very gradual rolloff), the effect at normal viewing distance was nearly indistinguishable.
As part of this study, I determined that part of the problem with achieving good I/Q results was that the transmitter Q filter was under-specified by the NTSC. In order to completely avoid the sideband crosstalk, it should have had a trap at 910 kHz, corresponding to the sound carrier offset. then the lower Q sideband would be curtailed the same as the upper sideband. Without this feature, some encoder Q filters that met the NTSC spec still would produce significant energy at 910 kHz, and this would be visible in the I channel.
Another way to avoid this crosstalk that cannot be done in a 6 Mhz channel is to have the full upper chroma sideband available. You can do this today if you have an I/Q set with a composite or S-video input fed from a wideband source, like a DVD player. In the lab experiments I did, we went through the whole sequence of comparing direct RGB to baseband YIQ to composite YIQ, to RF modulated YIQ to determine at what points the degradation occurred.
Another thing we determined is that even proper YIQ wasn't totally artifact-free. The I channel lies along flesh tones, so edges of faces are not distorted in hue when the color along the edge consists only of I -since ther is no Q in the flesh tone to begin with. But sharp-edged objects of other colors, like yellow letters in movie titles, would develop reddish-orange edges. This could be somewhat objectionable unless you were viewing from a good distance. The comparable effect in an equiband set is a reduction in color saturation near the edge, which is less noticeable unless you have the original RGB picture to compare to.
The one thing that definitely improved with proper I/Q demodulation is a reduction in red smear and improvement of the detail in red textured objects. My opinion is that the wideband I was wasted on the 15GP22 - the picture was just too small - but on a bright, modern 27 inch tube, say, it could have been quite noticeable, if not for all the problems of doing it well.
OP/ED ends....