If you've looked at the TMS9929A datasheet, you may have noticed it already outputs Y, R-Y and B-Y signals. But if you tried to hook these up to a component television, you may have noticed the blues are too strong and the reds are too weak.
The reason is due to the way television detect the black-level of the signal, and the way the TMS9929A designers assumed their device would be used. This picture is from the datasheet. Figure 5-10 shows the Y channel's timing. The sync pulse is the first negative going pulse to the left. This synchronises the television to the incoming video signal. Immediately next to that is the 'back porch'. This is assumed to be at black level and the television samples this point for reference.
Figure 5-11 shows a pulse on the back porch for the R-Y and B-Y signals. This is intended to activate the colour burst on a PAL modulator. Unfortunately, if you connect this up to a component television, it will sample that point and treat it as the black level (or rather the 'no colour' level). Because it isn't black, there will be an offset error which causes the faulty colours.
The trick with my circuit is to suppress the colour burst pulse.
Interesting! Many thanks for sharing this project!
If I understand well the schematic diagram, what you are doing is to take the level of U signal during sync pulses as voltage reference Vref. Then, you remove the color burst pulses by replacing them with Vref.
This works indeed, assuming U and V are at the same DC offset. Otherwise, the things get complicated. If voltage level of U and V during sync differs, you will remove the color burst from U, but not from V. In V you will have the voltage of U during sync. Which is also a pulse. Bigger than the original one. Which causes heavy color distortions.
I tried this in my lab, and I can tell the DC offsets of U and V do not match necessarily. And my oscilloscope is showing the color burst removal fails for V.
This could be fixed by generating another Vref for V. But that requires one more 4066 switch, and all of them are already used. In addition, Vref is used to subtract in some op-amps. My understanding is that this is used only to keep the output voltage into the 5V-0V limits. But having a new Vref adds a new question: which one to use in the op-amps?
I would really appreciate a clarification. And again, many thanks for sharing this!