Okay so here's the deal (I'll try to make it as simple as possible)
On the V2 SSDS3, the entire video section is corrupted by high frequency EMI (noise). It is proximity-based, meaning the closer you get to areas on the board with digital signals & power regulation, the worse the noise gets. The worst offending area is the mini-DIN itself, which is surrounded by digital planes, power, and middle layers of who knows what. The 4-layer design of the board is a major factor of this noise, because ALL 4 layers must be sectioned off and away from the video side of the board (which it isn't).
As proof of concept, I tested Voultar's internal bypass mod board "V.69" in my Super Grafx with all video LPF turned off (jumper set to unfiltered and OSSC video LPF turned off). This let me see just how much high frequency noise was present in the console with the SSDS3's video pins
completely removed from the expansion port. I then took jumper wires and tested the green channel on various
unpopulated areas of the SSDS3's video section to show how much (if any) actual high frequency noise is being generated by the SSDS3 from proximity contamination. Below are the results of this test:
Points taken from this test:
1. The left sample is with no wire connecting the console's expansion port green channel to the SSDS3. Note there is very little high frequency noise. Setting the OSSC to 9MHz LPF will cause this image to look a perfect solid green, which is why I turned LPF off in the first place so I can see the full effect of any noise introduced into the console.
2. The 2nd from the left image sample is connecting the green channel to an isolated pad on the
unpopulated video section of the SSDS3. Take note of how even with the pad being completely isolated on the SSDS3, there's still enough proximity EMI from the SSDS3's 4-layer design to actually 'infect' the isolated pad.
3. The 3rd sample is taken from the pin hole that the SSDS3's female expansion socket actually feeds into. Note how because this is a through-hole via, it passes through all 4 layers of the SSDS3. As such, the proximity EMI is even worse than the isolated pad.
4. Now look at the final image sample. This is taken from the trace that leads to the green pin of the mini-DIN socket. Keep in mind there are no shorted circuits. ALL of that noise is proximity EMI, and the reason it is so strong at the mini-DIN is because that mini-DIN is surrounded by digital ground planes, power regulation, and on top of that, all of the pins of the mini-DIN go clean through all 4 layers of the SSDS3 using through-hole soldering.
So what this means is not only is the SSDS3's video output contaminated with noise due to proximity EMI, but that very same contamination will actually crawl back inside your console and corrupt any RGB bypass modding you do to circumvent the noisy output of the SSDS3. Now the OSSC does a good job of hiding this noise as I said before using Video LPF set to 9MHz. But if you're on a Framemeister or some other device, you're going to see that noise show up in your internal bypass mods.
In conclusion, my regretful suggestion is to remove at the very least, the R, G, and B pins from ever reaching the SSDS3 so your internal bypass mod is not contaminated by that EMI. I myself spent over an hour desoldering the entire socket from the SSDS3 as seen here:
And then I removed the R, G, B, csync, comp video, and video ground pins before soldering the connector back onto the SSDS3. I now have clean internal RGB that is no longer contaminated by the SSDS3's EMI. Now this is a pain in the ass. So the quick (but more destructive) solution is to use flush cutters and snip the RGB pins on the SSDS3. That will be enough to keep your internal RGB lines free from contamination.
Lastly, this testing also brings bad news in that Voultar's FU-RGB board will be subject to that same contamination. The ONLY way to fix the contamination is to never allow those various video pins to ever reach the SSDS3. This also means the mini-DIN cannot be used as-is either. It must be completely removed from the SSDS3 and installed on an isolated board that has no connection to the SSDS3.