HDMI and RGB mods for different consoles

@OzOnE

How would you get on with consoles that are able to switch video modes or spdif output format/ encoding?

eg on the xbox v1.0-1.1 it has a conextant cx25871 has a serial input that you can set the register values for the video modes, raster values, flicker, h and vsync offset values,etc... there is a massive array of parameters crammed in there.

you would have to capture all this traffic and use it to set the hdmi output profile parameters and basically soft implement the encoder totally in the fpga? seems like a lot of work without a soft core to start from.

By chance is the data the same that comes out of the xgpu reguardless of video format the same I don't really know if it is dependant on the xbox output format ?

I believe the input is just multiplexed rgb on 3 seperate busses each 4 bit wide bus + sync and blank and the serial port.

 

Ah ok. Now I'm really excited for the Dreamcast version. No more crappy analog VGA boxes and I can finally use my DC on my HDTV instead of an old LCD PC monitor!

Ok, that would be really cool. Maybe someone could make a new CD add-on with HDMI out and a preferably faster CD-ROM drive.

I don't know much about the SuperGrafx, but I heard it is more powerful than the TG-16/PCE. Maybe somebody could make a SuperBooster that adds SuperGrafx capabilities? Someone get on that please!

 

Hi,

With most machines, it would be simple enough to detect the video mode purely based on the Hsync / Vsync timing.

You could monitor the I2C traffic to the DAC as well, but that's probably overkill.

The only thing you really have to set on the HDMI chip side is the VIC (Video Identification Code) for 480p / 720p etc.

I'm looking to add a small soft-core to the FPGA as well, since it has a fair bit of logic left atm.
That would primarily be for a simple on-screen display (if it's not too much work), but could be extended to handle I2C monitoring and other stuff.

The core pixel format (on almost all consoles that are worth targeting) should stay the same AFAIK.
It's normally just the timing that changes between modes.

I know that some of these DACs support multiple video input formats, but most of them will be just using the same standard (24-bit parallel RGB, or bt.601 / bt.656 etc.)

I definitely want to try the Xbox soon, because it's one of the first machines to handle proper HD modes natively, and has a fair bit of GPU power behind it.

I stuck a bigger HDD in my Xbox last year and loaded it up with CoinOps 6, so that should look great via HDMI.

@MonkeyBoyJoey / duskdawn - Yep, the HDMI board is already taking the I2S digital audio direct from the DC, and the HDMI chip can support S/PDIF audio as well.

The PS1 / PS2 is already in mind for the HDMI mod, and should work fine with digital audio as well.

OzOnE

 

Can't wait for the DC version and the Xbox version! I have the official Component cables for my Xbox so I will be able to do a side-by-side comparison.

I would ask about TOSLINK on the Xbox and PS2 versions but then I remembered that the Xbox has TOSLINK support on the AV Connector and the PS2 has a TOSLINK jack on the back of the console.

 

Hello here.
I'm way more impatient about N64/PS1 HDMI than Dreamcast. I don't feel that Dreamcast via VGA is so ugly.
@Ozone : i followed your work on gc-forever, about GC and N64 HDMI. This project getting closer to its end make me really happy
Please, tell me if i understood well : your HDMI board would be a "generic one", able to fit in many different system ? Or will it be one for each console ?
Keep on the good work !

 

I am looking at some disassembly pics of the Dreamcast. It seems the CD drive still needs to be placed above the video board. Does it still fit then?

 

With your board you are using a hdmi transmitter why dont you make a hdmi tmds output in the logic of the fpga with a logic level buffer would it not be cheaper doing it that way than an ASIC?

Or is this to do with the license for the use of hdmi ?

 

He is using an Altera FPGA. It does not have TMDS outputs. You can use LVDS, but that has different voltage levels and it will not work with every TV. Furthermore, the hdmi transmitter enables you to add audio easily and should work with any TV as long as the video timings you supply are correct and supported. I am not sure if the voltages are correct for the tmds interface on Xilinx FPGAs.

One other advantage of using a separate chip is that it is possible to do high resolutions such as 720p/1080p. Afaik license is not covered by using a HDMI transmitter chip. I wonder how Retroactive is going to do this.

 

Both LVDS and TMDS are current driven, so the source-side current sources/sinks determine the voltage level at the load. LVDS is a horrible substitute for TMDS though since its completely incompatible electrically. The normal LVDS configuration driving TMDS inputs would ask the FPGA's output drivers to sink 66 mA / pair!

Xilinx's TMDS mode sort of has the correct voltage across the load--it sinks 10 mA / pair = 1 Vpp differential. But fixed 10 mA means you can't source-terminate the line for SI without halving the level, so that really sucks. Still, low-end Xilinx I/O can toggle at 800 MHz so 720p/1080i HDMI is essentially "free" and apparently reliable with typical cable lengths.

Anyways you don't actually need TMDS outputs, or even CML to drive HDMI, you can drive it with standard VML + resistive impedance matching network, but it wont be robust against non-standard input terminations. For 480p on Altera chips I'd seriously consider it.

 

I did not look up the chip doc's but I know the Spartan-6 FPGAs have tmds outputs asuming many modern ones do and I was wondering why it was not done down that path, as it gets you away from legal issuies, as if someone at home reprograms the fpga to output hdmi its not his fault

As OzOnE is just using a hdmi connector as a slim dvi connector to fit in his console between the dvd and motherboard

 

Erm, where do I start? lol

@MoockyLoock - The board will be generic, and will now have three solder points on it for configuring the FPGA for the different video input formats.

Apart from the N64, DC, GC, GG, and a few others, most machines are pretty much using 24-bit (or less) parallel RGB, so decoding for all the formats should fit into the SPI Flash for the FPGA. Then, all we'd really need to do for each system is bridge the appropriate solder points, which saves us having to re-program the board for each system.

@bagheera - Yep, the board fits perfectly underneath the GD drive, but you do need to remove the lower shielding on the drive itself.
There are a few variations of the GD drive board of course, but I have most of them (apart from a VA0), and it looks like they will still have plenty of room underneath.

@mrgreedy98 - I did look into using LVDS outputs to drive HDMI directly, or using an external chip to output proper LVDS levels.

I've also seen a few other projects (FPGA NES clone etc.) using a TMDS level-translator chip, but with the price of the HDMI chip getting slightly lower every few months, I thought it would be far better and easier to go with the real thing rather than risk it not working on some fussy TVs / monitors.

It's not to hard to get basic HDMI output working using LVDS, but as bagheera and Calpis said, it's not easy to guarantee compatibility when converting to TMDS. It would take a long time (for me) to code too.

Using the proper HDMI Tx chip also takes care of a lot of the exact timings, colour-space conversion, audio packet insertion, standards-compliant VIC codes, hot-plug detection / interrupts etc etc.

I tried a direct hook-up from LVDS to HDMI a few years ago on the N64.
It did work OK on my Samsung LCD, but my Dell monitor refused to even come out of standby when the HDMI was plugged in (I was very lazy, and didn't even bother attempting to measure the exact voltage levels. lol)...

(the sparklies on the image were due to the ribbon cable from the N64 being too long, and from having no impedance matching whatsoever. I just wanted to do a quick test)...

https://www.youtube.com/watch?v=NCbkDgCIKRQ

To be fair, it did work quite well for the lower resolutions, but when you go higher than 480p / 576p the freq of TMDS clock needed is pretty high.
A poor old Cyc III would be working just on the edge for things like 720p, let alone 1080p. The HDMI chip takes care of that for us as well.

Somebody already beat me to the punch with the NeoGeo, but he's using a simple adapter from LVDS too...

https://www.youtube.com/watch?v=bTamCo2C6kg

The clever part of his code is that he added stereo audio encoding though, which I don't think I've ever seen in an open-source project before...

https://github.com/charcole/NeoGeoHDMI


I might have to look into the licensing issues if "they" bring it up, but I won't be using the "HDMI" logo or mentioning it in the installation guides.
We already have an amusing name in mind for the board which [RDC] came up with for the prototype, and we'll probably stick with that.

OzOnE.

 

There is another one which I found only recently with some creative googling, even though it's older than NeoGeoHDMI: https://gitorious.org/spartan-3a-dvi-hdmi-video-generator/spartan-3a-dvi-hdmi-video-generator - check the "hdmi-proto" branch

 

This would be exactly like using LVCMOS to drive a HDMI input, unless the output happens to have current-limited sinking. LVDS isn't just a "voltage" compatibility issue!

There's another upside: the HDMI fee is covered by the transmitter, so you don't have to worry about mentioning HDMI on your product.

It shouldn't be any different than guaranteeing compatibility through a transmitter chip, you're supposed to length match and control the impedance of traces in either case. It would be a good idea to at least measure the signals at a few hundred MHz for a sanity and first-order SI check.

??? It uses the S6's TMDS standard + passive adapter.

 

Im down for the DC board or 50+

 

Sorry - typo - I meant "TMDS levels / driver chip" of course. lol
(They do exist, for things like laptops etc.)

Ahh - thanks!

I was a tad worried about that, and was hoping I wouldn't have to pay anything if we simply called the board something else, and referred to the HDMI port as the "High Definition" port etc.
I will double-check that before releasing the kits.

I'm assuming the licensee is already Analog Devices then, and they already have to pay the yearly fee and royalties to implement HDMI on the chip?

http://www.semiconductorstore.com/blog/2014/licensing-costs-HDMI/654

Ohh, I forgot he was using a Xilinx chip.
The chips do have nice features compared to Altera, but I just can't get on with Xilinx ISE for various reasons (even the newer versions).


OzOnE.

 

Really good news indeed.
I'll need a bunch of your boards :]

 

"For example, if a cable or IC is sold to an Adopter who then includes it in a television subject to a royalty, then the cable or IC maker would not pay a royalty, and the television manufacturer would pay the royalty on the final product."

This line worries me.

I think he is using Altera. When I look at the repo there is an Altera project and in the .qsf file a Cyclone II EP2C5T144C8 is mentioned. Not that it matters much. It could easily be ported to Xilinx.

I am looking into making my own board for Dreamcast (and if it would fit for other consoles as well, PS1 would be an option probably and I have plenty lying around). I already have a board that fits inside an N64 that uses an ADV7513. I can easily redo the layout a bit so it will fit in a Dreamcast then. Making it should not be much more expensive than a good VGA cable for the Dreamcast.

 

Yeah, I'm wondering about the license stuff too.

There's obviously no way I could pay upwards of $5,000 to $10,000 a year, plus royalties atm.

I think kevtris mentioned something about this when talking about his "HDMI NES" boards.
I'll have to do more digging.

Good to hear about your own HDMI project using the ADV7513.
I hope you won't be too "competitive" right at this time. lol (only kidding)

What sort of FPGA are you using? Have you soldered BGA stuff at home before?


OzOnE

 

I talked with you on Youtube quite a lot. I am the guy with the triforce avatar. ;-)

I am not really looking into selling these. I am mainly doing it to learn something and to keep myself busy. Furthermore, my boards tend to break over time. Probably due to connectors being plugged in and out, my solder connections break I think. So I can not really sell anything unless I am sure of its quality. I might sell a few to friends, as I will have quite a lot of unused PCBs and ICs (a lot cheaper if you buy a few of them).

I use a Cyclone IV EP4CE6E22C8N (not that different from what you are using). I can get that cheap from China (obviously not a very trustworthy source). Currently I just flood the pins with solder and then remove it with solder wick at the end. This is not a very good way, as sometimes too much solder is removed or pins are slightly bended etc. I already ordered flux so I can try some drag soldering. I think this will improve everything quite a lot and will make my boards from breaking after a few weeks.

For my general board (which I will definitely use for Dreamcast) I think I will order a stencil and try hotplate soldering with a cheap cooker. From what I have read people are getting really good results with it. I would still need to solder some decoupling caps at the bottom of the boards by hand, but that is not hard at all.

I have never tried BGA. You would need at least four layers for that and I don't think most cheap Chinese are able to fabricate traces routed between BGA balls. I would really love to try BGA at some point, but considering the PCBs will be a lot more expensive (plus I could screw up the layout) I do not really want to try that out until I am better at designing.

 

Ohhhhhh. It's you. lol

Sorry. My memory isn't what it used to be, so even if you'd told me that before, I probably would have forgot by now. hehe

I would definitely recommend separate flux paste (or flux pen) for almost all soldering tbh (SMD and through-hole).
I've just bought another big 100g pot of Kingbo RMA-218 flux as it happens, and I've used Amtech and similar syringes for years now.

I find that solder wick rarely has enough of it's own flux in either, so always good to add a bit of paste first etc.
Flux just makes the solder flow so much better, and improves the look of manual solder joints too.

I was never a huge fan of solder wick until recently either, but I do use it for getting rid of solder bridges etc.
I think it's because I never had good luck with using it for through-hole de-soldering on delicate old retro boards, which is why I bought a proper de-soldering station a few years ago (Duratool / Zhongdi).

I use a Metcal MX500 station with a 2.5mm chisel tip for most stuff, as I've got used to using both the flat edge and the corners (for smaller stuff).
I first used Metcal stations at work in 2006, then again at the same place in 2011, and I've never looked back.

There are much cheaper stations available that would probably do just as good a job, but the Metcal is very nice.

Your solder tip should always look shiny and un-oxidized if possible. If not, it's probably the tip itself, or type of iron.
Too many people are using those giant 60W irons from (ex) Radioshack which have a shovel-tip, and a huge screw holding the tip in. (*cough benheck *cough. lol)

Yep, I've never tried BGA reflow at home either, but I'm considering modding a toaster oven for that purpose one day.
I would likely have a hard time just routing the board design though, as I've only ever done 2-layer designs (hence why [RDC] is helping me with this new layout.)

The new HDMI board has changed a bit since yesterday, but will give us a few nice extra options for fitting to awkward consoles like the GC.

It will also have the option for a really cool feature that will hopefully make portable modders cream in their pants.

OzOnE.