Woo! I’m glad you got this sorted.
The reason I provide the option of handling C-Sync attenuation on the board is because not all SNES “C-Sync” cables are built the same. A lot of these cables from a few years ago are strictly pass-through with zero attenuation network.
The C-Sync off of the SNES board is driven directly off of the ASIC, it’s intended for high-Z terminations. It’s strictly intended for TTL devices, or with cables that have a current limiting resistor. The C-Sync output off of the 1-CHIP ASIC is an open-collector. So if you plug a SNES into a 75term device in TTL mode, it’s going to sink current and pull things down. Short the TTL jumper for high-Z load devices or if your cable has a series resistor to mitigate current-sink. I’ve found that the buffered C-Sync state off of the S-RGB doesn’t totally agree with a few TTL dependent devices. After source termination, the sync drops to roughly 2.5vPP.
(Damn, I need to calibrate the old girl!)
Since you had no series resistor in your SCART cable, the C-Sync signal is sitting pretty high, displacing the slice level of the trigger range on the OSSC. With the TTL jumper “open”, C-Sync will live around 500mVPP. I don’t know where the OSSC slices, but the bigger the signal, the higher it lives. So your intermittent sync issue makes perfect sense here.
And of course, Video will live right around 700mVPP @ 75ohm impedance, source terminated. 🙂
SCART cable standards are non-existent as they’re entirely balkanized. I decided to add this little feature over a year ago to accommodate all of the various flavors.
All systems that are sent to me leave outputting logic-level, high-Z c-sync. Everyone should just sync on luma (to avoid chroma coupling) anyways.