Something bugs me a bit though. Aren't the right colors are supposed to be the ones you get with RGB? S-video should be closer to RGB than composite, no? I mean composite colors might be the ones we are used to, but are they really the right colors?
Bonknuts talked about this somewhere; essentially, the RGB pins from the Hu6260 video encoder chip haven't gone through some kind of colorspace adjustment that occurs internally. So actually, the raw RGB is technically kind of wrong colorspace-wise versus the Luma/Y that the same chip produces internally and outputs via pin 40 (
this applies to the Chroma as well). Search his account if you want details.
There is VERY little chances that I'll ever spend 3-4 hours building another CXA1645 svideo circuit for the PCE...
Yeah, unnecessary for numerous reasons, the one above and the obvious one as far as how easy and cheap this is! Some foreign chip is never going to match the properly produced/mixed Luma/Y that the Hu6260 chip was already producing. What happened is it appears that technical people discovered the RGB+Sync pins first and the Luma/Y pin was discovered much later... So that's why the early mods required some foreign chip and hooking it up to the RGB+Sync lines. Luma/Y is produced by mixing RGB+Sync, 4 signals, and it's far more likely that Hudson engineers did a better job of mixing it properly than what some foreign chip can do by starting over, etc.
Still, there are more tweaks possible. You can get a pure/perfect Chroma by disconnecting the Luma/Y from the internal amplifier instead of using a filter capacitor - go a couple of pages back on this thread, I already talked about it. That's if you're willing to do that and restore Composite support by alternate means. The other thing is that the S8050 or 2n3904 transistors aren't the best way to amplify a video signal... There are chips that do a better job than a transistor and give you different/better results like the NJM2267 chip for example. Refer to
ConHuevos here who used my SNES Component amp circuit which I designed around the S8050. He built that first, but then bought the NJM2267 and built a far more complex circuit just for purely amplification purposes here (NOT mixing). It's visibly better than the result with the S8050 transistor (
which is equivalent to the 2n3904).
Anyhow, the simple point is you're dealing with a natively produced Luma/Y signal here and all you need is a good way to amplify it; going as far as getting another chip to mix it from raw RGB+Sync is never going to be as good and it might even introduce incompatibility problems as well (
because of the Sync mixing and amplifying!!)... But on the other hand, when it comes to amplifying, these chips will tend to be better than generic transistors, etc. Overall, this transistor-based S-Video circuit is very cheap and very easy, so most people will be able to do it themselves and that outweighs most other factors. If you're a perfectionist like me though, you'll wanna go further.
http://quebecgamers.com/sensato/Workshop/PCE_svid.htm
Great job on this BTW; a Javascript mouseover to switch between the S-Video shot and back was a good idea. My Turbo Duo Composite was pretty shitty I always thought, I think your shots capture it, even though it's with a capture card.