Debugging Atari 2600 S-Video Mod

Why am I trying to mod Atari 2600?

My friend Rich had 2 Atari 2600 (4 switch model, circa 1980) in his garage (which I call "museum") that we stumbled upon (I forget why we were messing around in his garage).  He said he doesn't play it any more because he threw away the analog TV.  Apparently, Atari 2600 output its video over VHF channel 3 or 4 (there is a toggle button on the console), and he--later than the rest of us--has  finally thrown out his VHF TV.  Because I like taking thing apart, I offered to "fix it" for him WITHOUT knowing anything about Atari 2600 or VHF; I don't even play video games.

Dorking with an S-Video generator circuit

After some Googling, I ordered the parts listed on this website (except FMS6400, which is obsolete; I should have ordered FMS6410 from Digikey, but instead I ordered FMS6410BCS-ND, which is pin compatible with FSM6400, but in hind sight was a mistake, as it is too small to solder well), and started bread-boarding.

When I connected the S-Video cable to the TV, of course things don't work.  So this post is about the debugging my breadboard.  The circuit (should) roughly works like this:

1. Multi-channel buffer to mix the outputs from Atari TIA chip pins 5, 7, and 8 into the base junction of a transistor.
2. Addition of transisot emitter, TIA 6 buffered output, and TIA 9 are sent to the CIN (chroma?) of FMS6400.
3. Transistor's emitter output drives YIN (luma?) of FMS6400.  
4. FMS6400's YOUT and COUT are sent to S-Video, and optionally, the CYOUT is sent to the composite video cable (the yellow RCA cable).
5. TIA12 and TIA13 should be the stereo audio output, which can be connected straight to (with a 1.8 K pull-up resistor) to the RCA audio cables.

After checking the signals and concluding that they are semi-alive going into the FMS6400 chip, I hypothesized that I might have the wrong S-Video generation chip.  This was a bit of a leap, since audio does NOT go through FMS6400 and yet, I was not hearing anything.  The gamble paid off: after getting a new FMS6400 alternative (FMS6410BCS-ND) and carefully checking all traces, I started seeing blurred image of the "Space Invader" and heard some noise.  It looked like the following:

I was stuck on this until I tried the same breadboard on another Atari 2600, which happened to be of type B.  I actually did NOT know there were different types of Atari until I started to look for other mods besides the S-Video, and landed upon this page.  Picture looks beautiful now, especially on Rich's big projector (my iPhone4 camera takes a poor picture in low light)!

Overall, the project cost me only ~$50 in parts, but A LOT of time understanding analog circuits.  But as Master Card would say, the whole experience was priceless.  Merry Christmas, Rich!

Appendix: S-Video

After checking that I have 5 V from the Atari motherboard (tapped from TIA20, VDD), I checked whether I have signals on TIA pins 2, 5, 7, 8, 6, 9.  Get the data sheet on 8201 C010444D-03?  "Longhorn engineer" actually used 8222 C010444D-03 in his screenshots.  The pinout for C0104444D - NTSC is at atarimax.com:

• TIA2: SYNC
• Luma
• TIA5: LUM1
• TIA7: LUM2
• TIA8: LUM0
• Colors
• TIA6: BLK
• TIA9: COL
• Audio
• TIA12: AUDI
• TIA13: AUDO
• Of potential interest
• TIA11: OSC
• Chip selects: TIA21~TIA24: nCS3, nCS2, CS1, CS0

On a (cheapo) logic analyzer, I see buffer output on TIA2, TIA5, TIA8, TIA6 during the power-up, like this.

This would mean SYNC, LUM1, LUM0, and BLK are active, but LUM2 and COL are inactive. TIA6 and TIA9 together make up the chroma, so the remaining question is whether missing LUM2 and COL is OK for forming a valid S-video image.  I guess I'll find out soon enough.

The transistor's base is the mixed (with heuristically decided resistor values) the buffered BLK, LUM1, and LUM2 (which seems to be low all the time, so what's the point?).  If I don't pull the base up to 5 V, the collected signals get filtered out, as you can see below.

With the pull-up of the transistor base to 5 V, and the voltage dividers filled out, the transistor base and emitter lines look line this (yellow: base, green: base).

Note that the signal amplitude (even the emitter output) is almost completely dominated by the SYNC (yellow below):

It's possible that the resistors from the buffer out to the base for the different channels have to be modified later for the best color.

Appendix: Audio

I cannot hear anything on the TV either.  Most of the audio signal happens on TIA13 (AUDO), but sometimes, there is a large signal on TIA12 (AUDI) as well,