Possibility of building a retro arcade machine WITHOUT a Raspberry Pi

[xanthrou]

How can you build an old-school arcade machine using parts from old 1980s-1990s hardware (Motorola 68k, YM2612, 486, Z80, YM3812, SID, etc.), like making a modern rival of Neo-Geo and Capcom CP? (in a similar fashion to how the X16 was built)

Are FPGAs a good alternative? (a la c256 Foenix)

[SlithyMatt]

It's hard to beat FPGA on price, even if availability is currently hampered. The MISTer has cores for arcade platforms, and you can play precise recreations of those games with the original ROMs. But if you want to roll your own, specifically for an arcade cabinet, and FPGA is still going to be the most cost effective, especially if you want to implement chips that are out of production, like the 68000 and the SID. But really, the way to go is a Raspberry Pi, which you can use to connect custom controls and output HDMI video and audio to a modern display, like just a cheap LCD TV mounted in the cabinet. Then you can emulate whatever you want with software, or use modern frameworks and tools to develop new games. To design your own board and populate it with discrete, hard-wired chips is a giant integration and acquisition hassle, as evidenced by the X16 project, and then you are stuck with a single platform and no software for it.

[TomXP411]

On 12/7/2021 at 9:01 AM, xanthrou said:

How can you build an old-school arcade machine using parts from old 1980s-1990s hardware (Motorola 68k, YM2612, 486, Z80, YM3812, SID, etc.), like making a modern rival of Neo-Geo and Capcom CP? (in a similar fashion to how the X16 was built)

Every arcade cabinet was different. So you'd have to build the board for a specific game and load it up with the correct ROMs. Building those boards takes time, money, and skill, and it's possible the parts aren't even available any more. 

On 12/7/2021 at 9:01 AM, xanthrou said:

Are FPGAs a good alternative? (a la c256 Foenix)

Yes, either as a partial or total replacement for parts that can't be purchased new or reliably purchased used. If you go with a MiSTer type solution and build the entire thing into the FPGA, then you can even run multiple cores on a single system. 

I would never even think about building a single, dedicated arcade board at this point in time.. if I did want to build a stand-up cabinet, I'd absolutely build it with either MiSTer or RetroArch in mind. Probably both. I wouldn't use a Pi, though. I'd use a decent mini PC, either a Micro-ATX system or something like a NUC Hades Canyon, with a decent GPU on board. 

 

[xanthrou]

What do y'all think,@Strider @BruceMcF @rje @Cyber @Scott Robison @ZeroByte ?

[Scott Robison]

I agree with the assessment of using FPGA to replicate the functionality of retro chips. I wouldn't mind mixing FPGA with real chips where real chips are still available, but then you have to think about what you mean by "real chips". If you want to replicate a C64 (just for example) with modern parts, you can't source a new 6510, VIC II, SID. There are some options available for some chips that people have already created, and those are great if you have an existing C64 and need to replace a single failed chip, but if you're going to create something new on a reproduction motherboard, for example, I think it makes more sense to get a single FPGA and just build the functionality of all the chips into it, rather than setting up a separate FPGA / ARM / whatever chip to replicate individual chips. And once you do something like that, it is a relatively simple matter to load separate bitstreams to replicate a far broader range of systems (like MiSTer).

RetroPi isn't a bad solution depending on what you want to play, though I agree with TomXP411 about getting something a little beefier. My success with RetroPi was limited and I didn't have the patience to stick with it.

[BruceMcF]

I think my grandkid's neo-arcade cabinet has Galaxian and seven other games that he calls "pixelated", and since it has Galaxian, I'm cool with that.

I assume it's a cheap SOC in there running an emulator, and not some fancy cycle exact FPGA simulator of the originals.

[Cyber]

I agree with Tom and Scott.

I want to add that cabinets were so very different, that some of them used very specific chips and hardware. Finding some of these components nowdays is very hard or close to impossible. Replicating some rare chips in FPGA also is not an option, because there is no image for such chips to load into FPGA.

I can't see any real reason in creating exact replications of these boards. May be only if you have a strong wish to replicate some specific board, and willing to search information about it and overcome obstacles on the way.

So using FPGA and implementing designs of known boards might be the best option. May be start with older and simpler ones.

Also worth no note that most of the cabintes were designed to run only one specific game, but some cabinets were designed to run several different games on same hardware. May be this will help in choice of the cabinet to get better satisfaction in the end. 

[Cyber]

On 12/12/2021 at 5:48 AM, BruceMcF said:

I think my grandkid's neo-arcade cabinet has Galaxian and seven other games that he calls "pixelated", and since it has Galaxian, I'm cool with that.



I assume it's a cheap SOC in there running an emulator, and not some fancy cycle exact FPGA simulator of the originals.

There is a pub in my city that has two arcade cabinets with about 20 games in each. You can play them for free. There are also modern game consoles in the pub like Playstation, so demand for cabinets is not high. And not all visitors play, most of them drink beer and eat snacks.

I like playing on these cabinets. Of course they use emulation to achive result, so they not even close to replicating real stuff. But what is really great is that these cabinets are installed in real pub with real visitors - this gives great feeling.

[rje]

If it were a labor of love, and you had the means (skills and component access), the motive, and opportunity, then targeting one particular game makes sense.  FPGA and CPLD (etc) makes your job easier, again assuming you have the materials and the skill.

In other words, you have to have a goal in mind that hasn't already been done.  "Because I haven't done it" can be a valid reason.... but it really does have to be your goal.

 

[ZeroByte]

A friend and I made a MAME cabinet several years back. We used a PC motherboard with custom cards and boards from this site called Ultimarc (not sure if it's still a thing) - there was a VGA card that had arcade-friendly VGA timings built in so it could run at native 320x200, 320x240, etc and you'd connect the real arcade CRT to it. Then there were these boards that connected to the button/joystick swithces and presented them to the computer as a PS/2 keyboard, and another one like that to do the trackballs.

The one thing about this design that kind of annoyed me was the fact that the monitor's vertical timings had to be adjusted for certain resolutions - because real CRT. In a dedicated cabinet, the CRT would be adjusted once and that'd be it, but in something like MAME that changes timings between games, you may have to reach in there and adjust the vertical size. But the image was indeed arcade-perfect, as it had real phosphors on a real CRT.