I was wondering, has anyone looked into using the docs from http://gamesx.com/wiki/doku.php?id=av:nintendodigitalav to use an FPGA to convert it into HDMI, or at least DVI?
As far as the link you provided I understand, one would first recreate what the IC really accomplish. To a new video format would be possible, but what is the win ? From pinouts.ru about the video connector on the wii what I read here is that the wii has no digital connector. So no real win here with an FPGA for wii. GC maybe, but is HDMI or Dvi any win for its 3d graphix ? Maybe I dont know my wii/gc things, the Macronix CMPV-DOL is still unkown (what it do precisely ?).
The Macronix chip converts from digital video into analog YCbCr or RGB. From what I understand that same chip is integrated into the mainboard on a Wii, so you could get the same signals from inside the console. A good first step would be to do it on GameCube, though, so it can be tested and debugged without modifying the console. So the idea was to connect an FPGA to the Nintendo Digital AV on a GameCube and read the digital video and audio, and then repackage it into a DVI (video only) or HDMI (video and audio) compliant signal. Xilinx has an application note and reference design for this: xapp460. The advantage for this would be less noise than the analog signal, especially with long cables or a switchbox between console and monitor.
Its not the same chip. The AV encoder in the Wii is different to the one in the GC cable. However, the rest should still apply...
Surely much easier to get a Wii and just get a VDIGI VD-W3 (as it is the only one that does upscale the picture well with nearly no lag) to make Gamecube games run in HDMI (at least that is what I do).
It wouldn't exactly be hard to decode the GC video, with 8 data lines it's appears to be straight-forward 4:2:2 Y'CbCr pumped at 54 MHz ready to be decompressed and TMDS encoded. Audio looks like I2S or variant, very simple. Unfortunately there is a hard part... the hardware. It's not easy to design a board for a recent FPGA with differential I/O, nor is it easy to populate the board. If you just want to build one of these for fun a pricey dev kit with the PHY implemented and a vendor HDMI IP core is the way to go. As for Wii, you can already purchase this: http://www.neoya.com/shop/wii2hdmi/ which claims to be digital (somewhat ambiguously) so apparently the Wii DOES put out DV. Exactly how it's enabled does seem to be undocumented online (not surprising). I'm interested, but not enough to put down any money for a Wii project right now.
From the Xilinx appnote it sounded like you can drive HDMI/DVI directly from the FPGA output pins (as long as you don't need to be 100% standard compliant), and their first experiment was with a HDMI cable cut in half. I don't care about resolutions above 480p so the resources of a smaller Spartan 3a board may be enough. I'm planning to get a FPGA development board anyway at some point, so a prototype shouldn't be impossible. If it works it may even motivate someone to make a proper board with just the minimum number of components to make it work, especially if the DAC pinouts/protocols for other consoles get decoded. So far I only know this for GameCube and N64. The Xilinx appnote even contains the code for a transmitter and receiver. It doesn't support all the extra gubbins you would need for standard compliance, but I guess enough to display a picture. I've got a few other projects to finish first, though. I'll wait for someone to crack it open and look at the signals before I believe this.
No digital video from the Wii video output. It's just a stupid component video to hdmi converter. Only solution is to grab it from the pcb inside.
The only thing I could see this being useful for is LCD/plasma lag. ADC's in most lcd's tend to suck and make classic games impossible, whilst HDMI games are snappy and frag-o-licious. I would be interested in this if it would cut down on lag time, from button press to onscreen reaction. But I doubt adding another converter on top will improve this.
You can, provided your FPGA has TMDS as a built-in differential I/O standard. I think this is limited to only the newest Xilinx devices which are anything but hobbyist-friendly. Provided you can source a suitable FPGA inexpensively, it's still probably not worth while without expert design help. Also what good is a design relying on vendor IP and hardened macro blocks? At that point your design isn't really yours and isn't even upwardly portable. Of course it doesn't matter for a one-off, but for a one-off you can use a dev kit. The cheapest I've seen is the $350 Atlys. A more universal design would require general purpose serial transceiver (fast PLL and shift register with differential I/O), which I'm not sure even exist, and a discrete PHY. In the GC/Wii's case, a 27 MHz pixel clock only means a 270 Mbps transceiver which is probably sufficiently realizable in standard logic. ----------------- Are we POSITIVE Wii cannot output digital video? It seems strange to not include it since pins can be multiplexed and this device is so cheap, and nobody has mentioned noise in reviews.
The Spartan 3a from the appnote starts ~7 GBP at Farnell, the cheapest board with a Spartan 3e (which I don't know if it has TMDS compatible I/O) is ~50 GBP. And there is even the option to use a dedicated HDMI or DVI transmitter chip, for example http://www.ti.com/product/tfp410, which means the design can be reduced to nothing more than a FIFO between the different interfaces. And FPGA design and VHDL isn't the rocket science it used to be any more. Of course, if you want to do a complete system on a chip, but not if you just need a bit of I/O logic and a bit of internal state. The design may be even simple enough to fit into smaller devices. Anyway, maybe I'll try it if nobody else is working on it and I have some time left over.
long thread, but basically someone was able to source some of those MX chips and build his own for Gamecube: http://www.gc-forever.com/forums/viewtopic.php?f=26&t=770 I suspect he might know something about the video signals?
Right we're over-thinking it. HDMI means no DAC, ADC, or MX chip necessary. GC already outputs digital of some kind. Nintendo naming the port 'digital AV' was not a mistake or misnomer.
It does not, unless you're going to adapt the LVDS outputs. Well, not that specific chip, it doesn't seem to handle audio. Of course this design is simple, but that doesn't negate the fact that designing a PCB for a recent FPGA is hard hence the need for an expert. VHDL isn't rocket science? Wonder why all the space cases prefer it.. By all means stay optimistic, you know what you're doing so I'll leave the thread alone.
I found a reasonably cheap Spartan3A board... though it is more of a breakout board than an development kit. It basically contains only a 200k gates FPGA, a PIC to program it over USB, and 4MB DRAM. The 200k gates are probably overkill, and unless for screenshots and upscaling you won't need much DRAM. I ordered two and I'll report back. They seem to be quite nice to play around with in general. (BTW, does anyone know if the N64 uses 3V logic, or if it is still 5V?) Has anyone looked at the AV encoder on Wii? Is it only an upgraded version of the Macronix chip, or does it have a significantly different interface? The only thing I could think of right now is that the Wii supports widescreen signalling.
N64 is 3.3V CMOS. Wii only supports anamorphic widescreen so there's no physical difference between 16:9 and 4:3 modes, the game simply renders to 640x360 of 640x480.
I know, though I'm sure the video format is automatically signalled over scart to select the correct format on TV. I'll have to check... By the way, the Wii renders both 16:9 and 4:3 with 640x480, but with a different camera transform. The TV then expands the image horizontally. Update: I just checked, and at least over component it doesn't switch automatically.
?? This is what I wrote. Well that is the definition of anamorphic widescreen. The image is scaled vertically too or there'd still be letterboxing. ^corrected.