Sega Dreamcast HDMI Adapter coming soon ..

one thing at a time.... but FUCK YES DO THAT NEXT..... Virtual boy and maybe if at all possible clone one... because i cant seems to locate one that isnt over $200... but also the N64 idea sounds awesome... but could the console handle that or are you talking about a N64 clone?

 

I think he means like using two consoles together with a special program on a cart that creates the 3D effects, similar to super computers that use 200 PS3s.

 

I wouldn't make the comparison with supercomputers.

Basically he wants to run exactly the same game with the same exact controller input, in total synchronization, and have the camera angle slightly different in one of the two console. Getting the video output of each console to a different eye would provide stereoscopic n64. The timing required would be very intense though.

This would be awesome of course, but it might be simpler to try it in emulation first.

 

Ah ok. My bad. It sounds like an awesome idea that I would love to try one day.

 

Yep - what FG said. hehe

Many years ago under the MESS emulator, I made a very minor code change so that the camera in Mario 64 (and other games) would be shifted over slightly. This was done by forcing a certain value in RSP DMEM for the projection (Edit: "View?") matrix.

It didn't change the angle of the camera to point back towards the centre of the field-of-view / main objects though, but I'm sure that could be handled too.

So, it would mean modifying one copy of the cart image on one of the N64s to shift the camera perspective, then making sure both consoles stay exactly in sync.

Running both machines off the same physical clock and Reset signals is easy enough, but the obvious problem is that each console will be rendering a slightly different perspective / number of polys, so it's likely to go out-of-sync very quickly.

(Obviously the controller input would need to be handled as well, and shared between both boards.)

To stop that happening, the FPGA or MCU controlling both consoles would need some way of ensuring the RCPs have both finished drawing the same frame before letting them continue to the next frame. I think many RSP microcodes use the "SYNC_PIPE" GBI command when they have finished drawing a frame, and then trigger an interrupt so the CPU can start loading in the next set of data.

It might just be possible to rely on interrupts and control signals alone, since the rest of the signals are internal to the RCP / RSP / RDP.

Just a fun idea really, but definitely worth exploring again, especially since a lot of people have 3D capable TVs now.
Your idea of trying this on an emulator is a great idea, FG. Even though N64 emulation under MESS is slow atm, it would be a perfect way to test the theory.


btw, VGA out has already been done on the Virtual Boy using some great detective work by clever people like [Wuff] and Furrtek...

http://furrtek.free.fr/?a=vbtvout

I'd guessed that the LED array on the VB was driven in a shift-register fashion, as there are only a handful of signals on the ribbon cable.
Some excellent work by those guys though - it doesn't look like the simplest of things to figure out.

I don't have a Virtual Boy either, so I'll have to save that idea for a rainy day.

OzOnE.

 

@OzOnE i am curious to see the soldering inside the dreamcast. Do you have any pictures or video to share? Thanks!

 

ditto.... would love to see how this thing looks installed..

 

Erm, I'd rather not show pictures of the soldering tbh, as it's not exactly ideal.. lol
The mod on the first DCs wasn't made to look pretty.

Oh, OK then - It's basically as you would imagine - individual Kynar wires tinned, trimmed, and soldered directly to the legs of the video / audio DACs...

http://postimg.org/image/z08giv9gn/

That was the very first DC I modded (in a rush. lol), and one of video DAC legs got lifted slightly off the pad, so I was getting that weird yellowing on the early vids (due to a missing data bit that mainly lined up with the Blue channel data.)

The new HDMI board has 1.27mm-pitch pin headers instead of those tiny pads btw.
The longest row of IO pins is now along the "rear" edge of the board too, but there are still plenty of pins down the side edge to make it easier to mod consoles like the DC.

The 1.27mm holes will also allow us to use right-angled headers so the board can be more easily connected / disconnected if need be.

You could even stack or daisy-chain multiple HDMI boards for multi-TV output, or extra features.


On the next DC, I soldered it with the wires pointing straight up (parallel to the chip legs), and it was easy enough to then just to secure the wires onto the top of the DAC chip itself with a bit of tape.

Here is the second mod I did, which was used for the latest "100% working" vids...

http://postimg.org/image/4wqtgt9s3/

(Soldering the wires that way also made it easier because I'm right-handed, so I could hold the soldering iron in that hand while keeping each wire steady with the left hand. )

The twisted pair of wires is for the main pixel clock, and the left-most wire is just a Ground wire.
This is because the pixel clock is one of the more critical signals (running at 54MHz), so needs to be twisted with a Ground wire to ensure a good output image.

The data wires don't seem to be too fussy about shielding etc., but they should ideally be kept as short as possible before connecting to the HDMI board.

I also have...

• Three wires soldered to the PCM1725 audio DAC chip,
• Two wires soldered directly to pins 6 and 7 on the analog AV port (to force the DC to "VGA" / 480p without needing a VGA box plugged in. Eventually 480p mode will only be forced when the HDMI cable is plugged in.),
• One extra wire soldered to either end of filter FL407 (which is the Composite Sync signal for when I sort out the 240p->480p linedoubler),
• And of course the two power wires (+5V and Ground).

When the first big batch gets made, I need to ask Ronald if he can add some pads for +3V3 too. That will be needed for the buffer boards for the "5-Volt" consoles, and will also allow the HDMI board to be powered by 3V3 in machines where 5V isn't readily available.


You don't actually have to cut a hole in the metal shielding on the DC btw, as there's enough space in the gap where the screw hole is, near the AV port...

http://postimg.org/image/x9dg096eb/

(Obviously it would be good idea to add some foam protection for the wires, and ensure they don't get damaged when putting the screws back in.)

The ideal for the DC, Xbox, PS1, PS2, and other machines will be a QSB or proper Flex cable of course.
A custom flat Flex would be ideal for the DC, as it could then be passed between the gap in the metal shielding an AV port.
(the shielding may need a bit of bending to make sure the flat flex doesn't get damaged.)

Here is a paper mockup of what the QSB could look like. A custom flex would have the QSB part integrated into it too...

http://postimg.org/image/saf4y5825/

The QSB part would need to be thin enough to be able to solder from the DAC chip to it's pads.
You can get very thin PCBs made fairly cheaply, but a custom Flex would be more ideal in the long run.

The other-other option will be available on the final prototype HDMI board - it now has the FPC ribbon cable connector on the underside for extending the HDMI port. That means the whole board could fit underneath shielding etc., and be mounted directly onto the consoles' mobo.

The final board has a Mini-HDMI socket now too, but I could supply versions of the board with only the FPC connector fitted.
That would make the board even thinner, and easier to mount under shielding (like on the Gamecube).


I hope those photos don't put anyone off trying the manual soldering method?
This will only be for people who really can't wait for the QSB / Flex to be done, and as long as people understand that they may fry their DC in the process (of which I / we can not be held responsible. )

OzOnE.

 

thanks a lot, that was really instructive and... don t be ashamed of your soldering skills, they are great and very good idea for the QSB / flex ! i don t mind soldering lol but if qsb makes thing easier, i guess it would be worth the wait. but can i really wait? LMAO oh well...

Anyway, great job ! not much things to say, just it is incredible work ! can t wait to put into my DC ! cannot harder than changing a bios right?

 

Great job...I will buy one for sure

 

Thanks.

A ton of work still to do though. lol

Well, compared to a BIOS mod, it's mainly difficult because of having to use all the separate wires atm.
I might have to try some 0.635mm pitch ribbon cable next. The solid-core stuff might work quite well for the DC.

btw, the Game Boy Player arrived today - I'm currently trying to solder wires to it while sat on a bed. lol
I've mapped out all the test pad signals now, so shouldn't be too hard to get HDMI running on it.

Let's hope the custom chip which normally interfaces it to the GC won't interfere too much (like holding the /Reset line Low or something).

OzOnE.

 

Beautiful stuff, OzOne, I'm very interested in purchasing when available.

Thanks for your hard work.

-a2a

 

Oh well, you re on very good track to finish, probably the hardest is behind ! soldering wires while on bed? hey ! i am not alone after all lmao. Or on the floor, sometimes no light, sometimes no magnifying glass still wondering while it is working lmao ahh well...

i don t understand why going through the gameboy player, i thought u nailed already the GC, i should read again this thread more carefully.
thanks a lot !

 

He wants to bypass the gc and the required disc that is over priced

 

Yeah, I still haven't moved house, so no workbench atm. I also spent way too many years soldering on the floor until recently.
I've definitely damaged my eyes / back / neck / legs without the use of a proper workbench over the years, so be careful.

I did get the Gamecube working with HDMI before the Dreamcast (if you don't include my N64 -> FPGA -> Direct to HDMI experiments from years ago). Trying the mod on the Game Boy Player was just a fun exercise tbh, as I doubt it's ever been done before (not with a true digital connection at least.)

As LordFenixNX said, the fact that the disk is over-priced is another reason to try this...

You often see a fair few Game Boy Players on Fleabay going relatively cheaply, but without the disk. So, if / when they can be HDMI modded, it would be good to make use of the units in stand-alone mode and bypass the need for the GC + Disk altogether.

No luck just yet with the mod though - I've hooked up enough wires to start testing though.

I've also mapped out all of the test pad signals, and even managed to map out most of the pinout of the GBS-DOL custom chip (the one that interfaces the "CPU AGB" chip with the GC).

Most of that was thanks to a partial GBA schematic I found online, along with some other pinout info on the AGB chip. The rest was down to spending a couple of hours figuring stuff out using a multimeter and a lot of patience.

But, I haven't fully worked out how the power and clock signals are enabled just yet.
I've tried feeding in a 4.195MHz Master Clock signal into the AGB chip, and forcing the /RESET signal HIGH, but I haven't been able to get it to boot stand-alone yet.

Also, all my GCs are now in the storage container, so I need to confirm whether it does actually output 3V3 from the expansion port?
It looks like it does output 3V3, but ideally I need to confirm 100% that it's not 1V8 or something else.

(if anyone does happen to get chance to test this - I just need to measure the voltage on pin 1 or pin 35 of the GC's Hi-Speed Parallel port. Let me know if anyone can try that, but please don't break your GC in the process. )

I see now that the cart slot on the GBA / GBP has that small switch on the side for detecting whether a classic DMG, GBC, or GBA cart is inserted. The board then only switches to 5 Volts for the classic carts rather than 3V3 for GBA carts.

I vaguely recall the cart voltage stuff from when I owned GBA many years ago.

(Back in 2001, I was a bit bored at work one day, so I bought a Jap GBA on import from Lik-Sang via credit card using the work PC. lol)
(This one time....In band camp....)

Anywho, I'll get this GBP bitch to boot up eventually. hehe

Due to the video timing of the GB / GBA, I'll likely need to buffer a full frame on the FPGA itself, which means it'll probably only manage to fit 11-bit colour instead of 15-bit. It'll be plenty good enough for initial testing though.

OzOnE

 

oh yah good ! cuz i am pissed right now, i ve bought a game goy player a few month back, and didn t know it needed a start up disk to launch.. damm useless piece right now. price for the boot cd is way overpriced.

 

Not to worry now, guys - I just found a lone disassembled Gamecube still sat in the garage, so I can now check the voltages etc.

@acidtalk - Yep, quite a common issue. Sometimes the seller will post the original specs of the Game Boy Player which suggests that it includes the disk, and then there's not disk when it arrives.

Hopefully I'll have more news later about getting the GBP running.
I don't have the original PSU for this GC atm though, nor the (modded) disk drive, so I'll have to get creative. lol

EDIT: OK, confirmed the first part at least - the GC parallel port definitely has 3V3 going directly to pins 1 and 35 as expected. At least I probably haven't killed the GBP then. I didn't think they would rely on only 1.8 Volts on that port.

OzOnE

 

So can we expect a case that houses hdmi 1-2 controller ports and power supply that snaps down on the existing gbp case?

 

@LordFenixNC - probably not an actual case tbh, 'cos you can just install the HDMI board and extra controller ports into the GBP itself. It would need an external PSU of course, but that's easily done.

Also, a case that snaps on would need to work through the custom chip that normally interfaces to the Gamecube. That would be tons more work to reverse-engineer, and more pins etc. The FPGA bypasses all of that and simply connects to the CPU-AGB chip instead.

It's early days yet, but I've just had the first bit of success with the GBP - I found the signal which powers up the GBA chip now, and I can hear it booting.

I only have Wave Race for the classic GB, and Shrek 2 for the GBA, but both seem to be running OK.
I can see the pixel data whizzing past (via SignalTap / JTAG), and it all looks as expected.

You can see with GBA games it clocks out 240 pixels for each line, but with classic GB games it still clocks out the same number of pixels, but leaves a border either side so you only get the 160 pixels in the centre of the screen.

Since I'm generating the Master Clock from the FPGA too (usually 4.195MHz), I've also just played with overclocking / underclocking, and that works fine. The music and animations of course get faster too, but you can get quite a decent overclock by the looks of it.

Testing this has made me recall some discussions with [RDC] about whether we need an ADC for audio as well...

Some consoles / handhelds (like the GBA, GG, Lynx) still only output analog audio, even though digital video / LCD signals are available.
That means an ADC will be needed for those machines to be able to pass digital audio via HDMI / SPDIF.

That stuff can still be added on a small daughter-board, but it's reinforcing the idea that all of these machines could do with a custom QSB or Flex cable eventually.

Anywho, I'll keep plodding on with the Game Boy Player over the next few days.
I'll need to buffer the video lines before outputting them via HDMI, but only because the horizontal line period is around 73.44uS (13.616KHz).

It might just be possible to do a basic line-double instead without buffering a whole frame (no added lag).

OzOnE.

 

Fingers crossed. Keep us posted. Could your hdmi board work with a Sega genesis with 32x and CD ? Keep us posted. I already plan to send all of my consoles to a friend and have him install all the boards and required mods