#hardware design

So... in the broadest sense, and I think I've mentioned this before, I want to make this a 16-bit system (which is a term with a pretty murky definition), maybe 32-bit? And since I'm going to all this trouble I want to give my project here a little something extra the consoles from that era didn't have. And at the same time, I'd like to be able to act as a bridge for the sort of weirdos who are currently actively making new games for those systems to start working on this, on a level of "if you would do this on this console with this code, here's how you would do it on mine." This makes for a hell of a lot of research on my end, but trust me, it gets worse!

So let's talk about the main strengths of the 2D game consoles everyone knows and loves. Oh and just now while looking for some visual aids maybe I stumbled across this site, which is actually great as a sort of mid-level overview of all this stuff. Short version though-

The SNES (or Super Famicom) does what it does by way of a combination of really going all in on direct memory access, and particularly having a dedicated setup for doing so between scanlines, coupled with a bunch of dedicated graphical modes specialized for different use cases, and you know, that you can switch between partway through drawing a screen. And of course the feature everyone knows and loves where you can have one polygon and do all sorts of fun things with it.

The Genesis (or Megadrive) has an actual proper 16-bit processor instead of this weird upgraded 6502 like the SNES had for a scrapped backwards compatibility plan. It also had this frankly wacky design where they just kinda took the guts out of a Sega Master System and had them off to the side as a segregated system whose only real job is managing the sound chip, one of those good good Yamaha synths with that real distinct sound... oh and they also actually did have a backwards compatibility deal that just kinda used the audio side to emulate an SMS, basically.

The TurboGrafix-16 (or PC Engine) really just kinda went all-in on making its own custom CPU from scratch which...we'll get to that, and otherwise uh... it had some interesting stuff going on sound wise? I feel like the main thing it had going was getting in on CDs early but I'm not messing with optical drives and they're no longer a really great storage option anyway.

Then there's the Neo Geo... where what's going on under the good is just kind of A LOT. I don't have the same handy analysis ready to go on this one, but my understanding is it didn't really go in for a lot of nice streamlining tricks and just kinda powered through. Like it has no separation of background layers and sprites. It's just all sprites. Shove those raw numbers.

So what's the best of all worlds option here? I'd like to go with one of them nice speedy Motorolla processors. The 68000 the Genesis used is no longer manufactured though. The closest still-in-production equivalent would be the 68SEC000 family. Seems like they go for about $15 a pop, have a full 32-bit bus, low voltage, some support clock speeds like... three times what the Genesis did. It's overkill, but should remove any concerns I have about having a way higher resolution than the systems I'm jumping off from. I can also easily throw in some beefy RAM chips where I need.

I was also planning to just directly replicate the Genesis sound setup, weird as it is, but hit the slight hiccup that the Z80 was JUST discontinued, like a month or two ago. Pretty sure someone already has a clone of it, might use that.

Here's where everything comes to a screeching halt though. While the makers of all these systems were making contracts for custom processors to add a couple extra features in that I should be able to work around by just using newer descendant chips that have that built in, there really just is no off the shelf PPU that I'm aware of. EVERYONE back in the day had some custom ASIC (application-specific integrated circuit) chip made to assemble every frame of video before throwing it at the TV. Especially the SNES, with all its modes changing the logic there and the HDMA getting all up in those mode 7 effects. Which are again, something I definitely want to replicate here.

So one option here is... I design and order my own ASIC chips. I can probably just fit the entire system in one even? This however comes with two big problems. It's pricy. Real pricy. Don't think it's really practical if I'm not ordering in bulk and this is a project I assume has a really niche audience. Also, I mean, if I'm custom ordering a chip, I can't really rationalize having stuff I could cram in there for free sitting outside as separate costly chips, and hell, if it's all gonna be in one package I'm no longer making this an educational electronics kit/console, so I may as well just emulate the whole thing on like a raspberry pi for a tenth of the cost or something.

The other option is... I commit to even more work, and find a way to reverse engineer all the functionality I want out with some big array of custom ROMs and placeholder RAM and just kinda have my own multi-chip homebrew co-processors? Still PROBABLY cheaper than the ASIC solution and I guess not really making more research work for myself. It's just going to make for a bigger/more crowded motherboard or something.

Oh and I'm now looking at a 5V processor and making controllers compatible with a 10V system so I need to double check that all the components in those don't really care that much and maybe adjust things.

And then there's also FPGAs (field programmable gate arrays). Even more expensive than an ASIC, but the advantage is it's sort of a chip emulator and you can reflash it with something else. So if you're specifically in the MiSTer scene, I just host a file somewhere and you make the one you already have pretend to be this system. So... good news for those people but I still need to actually build something here.

It's a very fun change of pace to do something more in the vein of concept art than illustration. I haven't done this kind of visdev and item design stuff since back when I worked on AZAG