[Argus25]

....ps, the only way to assess the quality of the separated V sync is to disable the vertical oscillator (don't forget to extinguish the CRT beam current).

In any case, you can easily convince yourself that RCA was correct, if you sit down at a bench with just a TV pattern generator, a scope and a breadboard. Use the composite sync output from the generator and filter off the V sync with just one stage of RC filtering, looking with the scope. You will not get the result shown in post #71 of a properly filtered sync, it will be more like the result in post 90, with serrations, and the problem here is the the filtered pulse has the same amplitude on its rising edges at two different times, begging for interlace troubles. If you increase the capacitor value to eliminate these, the rising edge has a slow rate of change and the amplitude is attenuated and then, in use, the V osc is closer to threshold at the point of triggering and is more prone to be triggered into the next cycle by interfering H pulses.

Trying then adding more RC filter stages and it becomes obvious that this is how the pulse should be separated to attain a smooth fast rising edge free of serrations as shown on the image in post #71.

I could also add, if you have a TV and you find that the position of the vertical hold control affects the interlace, this is the inevitable result of a poorly separated V sync, with serrations on its rising edged and the start of the scan can be triggered at different times and is completely cured by a three stage filter, which ideally the set should have had in the first place, or a bare minimum two stage filter, of the correct time constant.