[old_tv_nut]

Looking better all the time - thanks for posting. I was surprised to hear that the IO response rolls off so strongly. I had always assumed that the output of the IO electron multiplier had only a small roll off due to stray C, like a photomultiplier tube circuit.

Is R20 (24k ohms) the IO output load? That combined with the stray capacitance must be the cause of the roll off. Of course, you need a certain size resistor to get the low frequency gain, but the capacitance rolls off the highs. In that case, the overall design is a trade off between needing more overall gain in later stages vs. more peaking (high frequency gain) in the later stages. The peaker circuit (R193 and C16) looks like it has a break point of 200 kHz. This would correspond to about 5 pF across 24k ohms, a reasonable sounding number. It also implies about a 20x boost at 4 MHz, not 80x.

--
The .05 coupling cap effect on low frequency response is not a problem if it is followed by a black level clamp somewhere further along. Is there a DC restorer (with a horizontal pulse) somewhere later in the circuit? There should be, otherwise the R, G, B black levels will vary with respect to each other.