Tested it on the CoreGrafx I mostly used and it has some jailbars and I have la steve fix on it which made it a little better, tried Tims, and made no difference.
Used the new Banding test,
https://dl.dropboxusercontent.com/u/1403041/banding.pceTested TG16 without any jailbar mods, and it was nearly invisible, tried a white PCE and it was not showing at all at stills, some less visible more like dreamy maybe when the orange moving squares was showing in the banding test, was struck by that those two didn't show anything when a double modded CG showed it....
But then again, 100nF should be enough on the rails, it's the most common capacitor value just because of the function it has, between a line or vcc to ground... and it's used for it's filter function
so the value actually has to be precise, but first you must know signalwise, what it is you want to remove, what we want to do is not much different from a speaker crossover but we don't know the "jailbars" frequency and not input or output impedance of the circuit we use, so we can't actually calcite on the crossover frequency that a specific capacitor value removes.
It's not like a capacitor in a power supply where bigger is better, it's a precise value making a crossover frequency that either makes a highpass or lowpass depending on how it's connected, and if the noise that this jailbar creates is on the same as the wanted signals frequency (RGB/Composite) it will never be solved, it might even mix internally of the HUC6260, and there's no fixing that...
But it's very different from model to model and between individual consoles that's for sure.
