Why can't people recreate retro console hardware?

[atheistgod1999]

The analogue NT was super expensive because they were taking hardware from actual famicoms. Why couldn't have they just manufactured their own PPUs? When I asked that on Facebook, everyone told me it would be just another clone console then. I don't get it. I thought clone consoles didn't recreate the original hardware and instead just emulated the console on a chip. If someone actually recreates the GPU/PPU (same pins and circuitry and everything), CPU, SPU, and everything else, it would basically be not a clone console but a perfect replica of the original, just like the NT. Can someone explain?

[Guspaz]

Clone consoles do recreate the original hardware, with varying degrees of accuracy. Emulation devices, like the RetroN 5, simply run pretty standard Android ARM boards with stolen emulators (in that Hyperkin stole the software they put in their device).

The problem with producing new copies of the original hardware is that manufacturing real custom chips is obscenely expensive, and the potential market is really small: when people buy stuff, they're usually just going to get the cheaper emulation boxes that are available at retail (like the aforementioned RetroN 5).

A middle ground is the use of FPGAs. Think of them like programmable processors: they're made up of thousands of tiny little elements, and you can program the chip electronically so that these elements are configured and connected in certain ways. In this manner, you can actually configure it to recreate different sorts of chips, and do it in hardware, not by emulation. You can program one of these chips, for example, to become an NES CPU and PPU... and that's exactly what some people have done. The downside of an FPGA is that they can't run at very high clockspeeds (usually just a few hundred megahertz) and they are much more expensive per unit than real hardware. But the advantage is that you can buy a few of them and do what you need without the massive startup costs to build real chips.

What I mean to say is, maybe if you wanted to make a real chip, it'd cost you hundreds of thousands of dollars in startup costs, but then you could make them at $1 a chip... Or you could buy a $50 FPGA chip that can recreate the same thing, but has zero startup costs. These are made-up numbers (well, really, FPGAs tend to cost around that much for what we're talking about, I think), but you get the idea.

Anyhow, one such example is the RetroUSB AVS, which re-implements the original NES hardware on an FGPA chip with high accuracy. It supports HDMI output, and is expected to cost somewhere between $150 and $200. That's basically what you're asking for.

You can take this farther, too: because FPGAs are reprogrammable, there is no reason why it can't simulate many different consoles. Kevin Horton (AKA Kevtris, creator of the Hi-Def NES HDMI kit) has been working on a multi-system FPGA-based console for years, and he plans to support pretty much all consoles from the 16-bit generation or older with a very high level of accuracy.

Why have more people not done this, you ask? Because it's a huge amount of work. Kevin Horton has been working on his for 10 years, and while the RetroUSB guys have only been trying to implement a single simple console (the NES), they're still a few years into the project and not quite ready to ship yet.

There are a variety of FPGA implementations of retro consoles out there, but the problem is that most of them are not very accurate (they're incomplete or buggy), and the boards they run on are not exactly cheap. For example, the MIST board that a lot of FPGA consoles run on, it costs around $220 USD, and doesn't have any built-in console controller ports or slots. It's a really hard business case to turn that stuff into something you could sell to customers. You need to either have a *REALLY* polished hardware/software setup like the RetroUSB AVS, or you need to support a large number of consoles very accurately, like Kevtris' work.

[telemetry]

How exactly can you replicate a retro console chip? Nintendo and Sony didn't publish VHDL (basically open source code) for the chips, so there's no way to know by casual inspection how to refabricate them.

Emulation works by knowing the logical behavior (ex, at an register/assembly language level), but it's far easier to test and verify emulation because all of the work is done in software.

[Voultar]

How exactly can you replicate a retro console chip? Nintendo and Sony didn't publish VHDL (basically open source code) for the chips, so there's no way to know by casual inspection how to refabricate them.

Emulation works by knowing the logical behavior (ex, at an register/assembly language level), but it's far easier to test and verify emulation because all of the work is done in software.

You can; Grab a litre of sulfuric acid, strip away the poly-silicon, take some nice polaroids with a high resolution microscope, take more photos, and a few more, and just a couple more, and begin replicating all of those small transistors/gates.

For the most part, it's all theoretical and academic in nature, with very little "hardware de-engineering" as briefly described above. That's why 95% NoaC's and other hardware discreet clones suck massively.

[atheistgod1999]

How exactly can you replicate a retro console chip? Nintendo and Sony didn't publish VHDL (basically open source code) for the chips, so there's no way to know by casual inspection how to refabricate them.

So why can't tech wizards deconstruct it and find out how they work?

[Elixir]

The analogue NT was super expensive because they were taking hardware from actual famicoms.

No.

[atheistgod1999]

I've been thinking: how do bootleg SNES minis work? Do they emulate or have copies of the original chips? I thought those came out way before the other clones. I also wonder why only that console has bootleg versions.

[Guspaz]

Some are just copies of the original chips (to varying degrees of accuracy), some are are just emulating. Most of these older chips are very well understood at this point, and when you're doing an FPGA implementation you're not recreating the chip at the transistor level, you're recreating it at the logic level: given a certain context, what does instruction XYZ do, and how long does it take to do it.

It's not like these chips are a mystery, the CPU in the SNES is a modified version of the processor used in the Apple IIGS, while the CPU in the NES is a modified version of the processor used in the Apple II. Those CPUs (the 65c816 and 6502 respectively) are well documented, so only the changes made by Ricoh on Nintendo's behalf are missing, and people figured that stuff out ages ago.

[panzeroceania]

atheistgod1999

it's simply too expensive. Even after you have decapped the chips and created a schematic to fabricate more, the fabrication process is too expensive, and there are not enough customers willing to pay full price for an old console.

1. fabricating an entirely new console is the ideal, but most expensive

2. next best is an emulator at the hardware level using an FPGA, this is a good solution as it still provides better results than an emulator on a general purpose computer, while still not the same as having an original system. These do exist on the market

3. the worst is emulators on a chip, typically an arm chip. This is an emulator running on a general purpose chip, this doesn't really provide any benefit over an emulator on your computer, in fact it is often less accurate than the best emulators available on computers. Sadly, this is the most commercially successful category on the market because it is the cheapest.

it all comes down to economics and the size of the target market.

[antron]

atheistgod1999

it's simply too expensive. Even after you have decapped the chips and created a schematic to fabricate more, the fabrication process is too expensive, and there are not enough customers willing to pay full price for an old console.

Someone needs to make a kickstarter to test this theory of being to expensive.

[kamiboy]

Ducats to donuts that Nintendo and their manufacturing partners could do it for cheap if they wanted to.

I think AV Famicoms had a very low retail price in the early to midt 90's just before they were take off of the market.

A lone hobbiest though, I do not think so.

[Guspaz]

Putting together a kickstarter is a fair amount of effort, and requires a working prototype. Do you really think that a kickstarter that involves paying the hundreds of thousands or millions of dollars required to get chip production up and running would succeed? An FPGA can recreate the hardware of a retro console perfectly down to the quirks of the composite video signal, if somebody puts enough effort into it. That's more than sufficient. The product will always be a clone, whether it uses real copies of the real chips, or FPGA copies of the real chips.

[mickcris]

there are plenty of working old consoles around for sale. no need to go making new ones really. Kevtis reverse engineered a lot of systems and is probably going to be releasing a system at some point. not sure if its going to have all the cartridge ports or just run roms

http://atariage.com/forums/topic/242970 ... me-system/

Some of these cores were maybe going to go into the Retro VGS (who had a horribly failed indegogo campaign). It had many issues and one of them is it was too expensive so people did not want to pay for it.