Neo Geo AES RGB Over Saturated

So this weekend I finally bought an AES system. Japanese mobo rev AES3-6. With composite it gives a perfect images (as perfect can be with composite anyway). After soldering myself an RGB cable, I was not as pleased. Crappy in three ways:

1) Jailbars,
2) No csynk,
3) Over saturated colors.

For 1) I just had to do the simple RGB mod. I tested all different versions including removing the caps and resistors, patching both sides but finally ended up in restoring it all and cut the traces. For 2) I patched directly from pin 10 of the CXA1145 to pin 7 of the video connector (had to cut the trace for pin 7 on the mobo, otherwise it was directly connected to pin 3, mixing up composite video in my sync).

After fixing 1) and 2), the picture is razor sharp with perfect sync. However, the colors are extremely over saturated. Almost like CGA graphics. It has nothing to do with brightness/contrast adjustment. Despite raping the mobo, composite is still working nicely, but I don't do composite. I demand perfection.

I've been looking at the schedule, but I don't find many possible error sources. If it was the 74LS05 or the 74LS273's, I suppose it would affect the composite video as well, no? I've measured all resistors and they seem fine. I've experimented with or without 75 ohm resistors on the R, G, B and even sync, but it only affects the brightness.

Anyone having any idea about possible reasons? I would be happy for any clueful suggestions.

 

You should use the resistors on RGB, never on sync. SNK were pretty poor at outputting the correct 75 ohm

The revision 6 board has crap RGB, as you found out.

Which mod did you use?

http://nfggames.com/games/neorgb/
http://www.mmmonkey.co.uk/snk-neo-geo-aes-rgb-bypass/
http://wolfsoft.de/wordpress/?p=247

 

I just cut the traces, like here: http://www.mmmonkey.co.uk/snk-neo-geo-aes-rgb-neo-aes3-6-vertical-lines/

However, as I mentioned above, I've tried all googlable variants, just to end up with the simplest one of them all. It's no difference between the results of them, basically.

Indeed the RGB is crap without modification, but at least I see now after the mod that the quality is perfect. It's just the extreme saturation that is the problem. I need to know what might be causing this.

I don't know really what those 21 resistors do. Their values are correct according to what is screenprinted onboard, but maybe they should be adjusted anyway in order to tune in a sane level of saturation?

I just tried another way. Now I've tapped from the R, G and B inputs of the CXA1145p directly to the output, only via 75 ohm resistors. The result is quite interesting. It is by far too dark. Can barely see anything. However, if I bring the contrast to the max and surf up the brightness until the picture is as good as gets, I see the colors are much better. My conclusion is that the CXA1145p is doing something bad to the signal...

Or, just a moment here... Looking at the datasheet for the CXA1145p, the recommended peripheral connections are different from the implementation of the mobo. In addition to what's on the mobo, the signals should be pulled to ground by 75 ohms, which is not happening. Is that the missing part of the story?

Nope. Didn't help.

Starting to suspect the CXA1145p chip. Maybe I should try sourcing a new one, or try bypassing that chip with transistors amplifying the RGB signals?

 

Merged your posts. Don't forget there's an edit button heh

A bad CXA1145 sounds unusual... with that result, anyway. I haven't been inside an AES in a long while, but they didn't really use the chip per the datasheet IIRC, so there's probably something you've taken out of their circuit that was necessary. MVS was even worse for standard televisions. You could try using a CXA1645 with the suggested circuit, perhaps. Whatever circuit you use, make sure the impedance is right. I'll have a look at the schematic when I get a minute.

What are you using as a monitor?

[EDIT]
Put the caps back. Diagram shows running through a 100u 6.3v electrolytic with the positive on the chip side, then a 68 ohm resistor. Try that first (breadboard it if you like) and see how it looks. If there's no colour improvement (you may still have a crap image), it's worth trying a different encoder.

By the way, those ground connectors on the datasheet are only to test the electrical characteristics. You could, of course, build that circuit to test the chip. The application diagram is further below and again has an electrolytic (470uF) and resistor (75 ohms).

 

It took 25 minutes to forget about the edit button

 

I somehow managed to quote myself instead of editing :\

 

Thanks!

The over saturation was present already from the beginning, before I even opened the console.

I tried bypassing the encoder with a transistor amplifier, but that did not help. The result is the same. So the encoder is likely not the problem. I find it quite strange the composite picture has correct colors, as if I understand correctly, it is encoded based on the RGB input to the CXA1145p. I must be overlooking something.

By the way, my board is a little different from the schematics. The caps C36-C39 are 470u and the resistors R62-R65 are 75 ohms. Also they've added ferrite beads after the resistors. I've tried restoring everything to original, but that does obviously not help.

I'm using a Sony PVM and a JVC reference monitor, both with a BNC connector per line. I'm also using an Extron RGB switcher for convenience, but of course I've also tried without that one.