[tech] Neogeo the mystery ?!

Decapping ICs is something for the rich/privileged few with access to such lab equipment and even then only used as a last resort. Even once you decap a chip, it's not remotely the easiest thing to understand, which is why usually only ROM is gleaned. This may be good for finding MCU/DSP instruction sets, or reading out some protected chip's fuse array, or even with crazy tools reattaching protection fuses, but there's no way someone smart will try to figure out a Neo Geo chip at the switch level looking at photos taken using a microscope with digital camera attachment when you can just look at algorithms in MAME and signals on a logic analyzer and put two and two together.

 

Someone has been decapping SID chips lately.
A few years difference though, so you do have a point

 

I still think that's a little crazy, something only for SID worshippers but there's a lot of them.. NES sound has been RE'd extraordinarily well through blackboxing, though I guess it's a tad simpler than SID and it's fancy analog junk

 

Especially on Neogeo ! There is a lot of consoles to emulate. All popular machines like SNES, GEN, PSX are allready emulated and almost fully functional with a lot of documentaion.

Even the Saturn, which is the one of the most dificult to emulate, but you can find a lot of sheet and information about saturn's chip.

Neogeo is not specific enough. It's a very standard arcade machine and yet she lacks a lot of informations about base knowledge like which chip handle sprite, layer, and real hardware sfx capabilities...

 

This is also the case with Neo Geo though..?

You can find information about Saturn chips? Where? I know you can find programming information from leaked Sega documents, and if SNK's Neo Geo documents were leaked too there would be a few things to learn, but really chip functions are empirical and I think the Neo Geo is 90%+ understood.

Chips can be understood by just looking at the PCB and which components the chips connect to. Can you get a big picture of the PCB?

Also it's entirely understood how Neo Geo graphics and sound work.. I'm not sure what more you're looking for. The few things that aren't understood are some unknown register bits for very minor features, and the low level operation of the glue chips and VDP, but such a thing doesn't nearly exist for Sega Saturn (or any Sega console except perhaps SMS AFAIK).

 

No it's not. A very simple example. Go to Neogeo' s Wikipedia page and read technical spec. Now, go read NES, SNES or Gen spec, you see the difference ? OK, Wikipedia is is the public domain of knowledge, It would mean that the Neogeo is not a part of this ?

For programming purpose, you're right, many consoles are well documented. But on hardware documentation ? When I open a Saturn I know where are VDPs, SPU work ram and vidéo ram... Same thing for PSX, Gen, SNES, PC-Engine... I know who is doing what.

Right again ! However, on Neogeo "just looking" does not work. What I see on NG PCB ? Z80 and 68k CPUs, YM2610 sound CPU, BIOS, memory chip include wRAM, vRAM, cRAM and pRAM. And yet, cRAM the pRAM are not documented on "standard specifications". But other chips ? Where is VDP ? Zom and scaling can realy render by hardware or it's only software ? Diference between PCB revision (merged chips for example)...

Another example : in later models of Saturn, where is the 68k (I know it's merged whith another chip) ?

What I looking for ? Like others consoles, I search to build clear and precise information about what each chip do. It's not for developement purpose, it's only for inform. These little tips, can help for hardware modidification (it's a use).

 

Just because it's not on Wikipedia doesn't mean it doesn't exist, most document writers probably don't want to contribute directly to Wikipedia or even know how.

But you don't REALLY know how it works inside, or exactly what each chip is doing without studying the bus. In the case of your examples, the consoles you mentioned all have very compact hardware implemented in few ASICs so it is very obvious which chips do what. In the Neo Geo, there are many ASICs which are less integrated, similar to "glue logic". The only way to understand what these chips do is try to reverse engineer them which isn't necessary on a SNES..

Well, if you only use Wikipedia for information then of course you won't figure it out. If you can upload a picture of the full motherboard, I will try to label chips.

PS: Scaling is hardware (there is even a scaling ROM on the PCB), "zoom" is software, but really it's just large sprites which are scaled to appear smaller.

Is the 68k connected to the cartridge bus? (I think so) So compare 68K cartridge signals from old model to new model and you will learn which chip contains the 68k.

Most people don't see this view, they only care what the chips do when they want to learn exactly how the chip works. Without that information you can't really do intelligent hardware modification anyways. Another thing to consider is that there are no Neo Geo schematics, even 3rd party or service manuals available online. All the other consoles have that leaked because of the large amount of 3rd party developers. Also because of the system complexity less people would attempt to create a schematic or reverse engineer the hardware. So there are reasons why this info doesn't exist, but it's no less your fault for not reverse engineering than anybody else

 

SID is worth it. It also has been emulated quite well without decapping, but going to for the last few .001% of accuracy is worth it considering the importance. I still love alot of SID tunes. Bob Yannes was a genius.

 

.001%? The SID is not even close to being as accurately emulated as that, and never will be. It is impossible to accurately replicate the analog side of it, not even MOS/CSG themselves could do so which is why no 2 SID's (even if they're same revision) sound exactly the same.

Couldn't agree more with your opinion on Bob. The SID is simply an incredible piece of sound equipment, especially considering it was designed in 1982.

 

Why can't MOS replicate it? I'm sure it must have once been prototyped in discrete logic and passives... Besides, when passives are on chip, they're more than precision trimmed, I don't know how you can hear the difference.