There’s a new RGB bypass board coming out soon for non-1Chip boards. If it isn’t the caps, perhaps that will fix the problem. Hopefully. =P

Real crappy highlighting, but here's the path to the multi-out. C47 isn't inline and Q7/Q8 (transistors) are both involved with the RGB encoder and the raw output. R19 would keep the line high, which should result in blue being solid bright on the RGB output? I'd double-check that first.