GBS 8200/8220 CFW Project

[rama]

culexus:
Glad it works well for you
The Xbox already works well in standard resolution, be it via RGB or vy Component input.
What I'm working on right now is support for enhanced definition 480p / 576p (PAL).
The benefit will be that 480p doesn't flicker, while regular low res in that console generation almost always uses interlacing and thus flickers.

[AndehX]

Ok. I assumed the direct capture would be more accurate to the original intended look. Maybe i'm wrong. The GBS doesn't look bad by any means, and on my projector is looks perfect. I'm just curious about the difference in perceived brightness.

[rama]

The brightness seems really low in the GBS capture. That's why I assume a problem with the GBS DAC to capture card input.
It's like there was an additional voltage divider or a non-standard termination on the capture card. Something that would drop the nominal 0.7Vpp GBS output to lower than that.
The GBS is meant to drive a VGA display, so it expects 75Ohm termination. If you check your card on the analog inputs, there should be the same 75Ohm resistors somewhere.

[Furtin72]

Hello there!
I‘m new to this forum and stumbled upon this thread after having finished a „Xbox360 to Jamma“ project ...

I just wanted to kindly ask if you have an idea why the picture is still somewhat blurry (from downscaling?) and have a slight flickering to it?
Maybe you can point me in the right direction...

I have made a video of my setup and it includes all the parts in the description text:
https://youtu.be/0RsSpQY5CSk

I‘m glad for every help!

[AndehX]

Got my PLCC sockets today, so here's a few pics of my setup now

[Higgy]

AndehX - nice work there.

Furtin72 - i see your not using a GBS82** board and there are hell of a lot of wires to go through for remote fault finding.
Can you remove all but the video wires to reduce the amount to fault find? You might also find a topic that uses that exact board to get people that know it. Good luck with it, looks fun. I thought about an XBOX bartop but used a Raspberry Pi 3 instead.

[rama]

AndehX:
Is the connection good? Can it withstand a slight knock to the board or plugging in a USB cable?

Furtin72:
I don't know what kind of hardware that is. Probably not a TVIA 5725, which this project is all about.

[AndehX]

AndehX:
Is the connection good? Can it withstand a slight knock to the board or plugging in a USB cable?.

Yeah this is probably the only issue. You can get it to fit snugly, but you have to push down very hard. Otherwise it just pops off. Once it's on properly it can take a knock or 2 I'd think. Plugging in the usb cable a few times will probably make it pop off tho.

[Higgy]

Just to say I am testing on a Sega Saturn and its looking good!
Colours are really vibrant.

[rama]

I'll order a few sockets so I can test it on a few different GBS boards.
If it's generally okay, this is probably better for end users than soldering wires.

[AndehX]

Could maybe knock up some premade ESP boards with sockets and sell them on ebay or something for those who dont have the skills to set it up themselves.

[Furtin72]

@higgy and @rama thanks guys. Didnt meant to derail this thread. Also, I found another solution and actually saw where the problems (interlaced/flickering) came from. I‘m using a proper downscaler now. Thanks nonetheless and happy easter!

[rama]

Not a problem, @Furtin72

Those other converters are interesting as well.
Who knows what kind of cheap chips are out there, waiting to be used to their full potential!

[Higgy]

Hi,

Just to say I got my D1 Mini today and wired it up. Very simple to wire up:

Connections for a WeMos D1 Mini are as follows:

D1 = SCL
D2 = SDA
G = Ground
D7 = VSync
D6 = Debug

5V = +5 from GBS82** power header
GND = Ground from GBS82** power header

Just been trying my Playstation 1 and it is looking good

[Syntax]

Is the GBS pretty much the best solution for CGA to VGA now?

Has anyone checked input lag?

[rama]

It depends what timings CGA presents. At this time, the software works with timings close to regular PAL or NTSC, as well as their progressive variants.
Anything way out of those specs won't be recognized and the software doesn't know what to do with it.
The chip itself probably supports it just fine, so it's a problem of adding new presets.

Lag is untested but will be around one frame, ie 16ms or 20ms, depending on the format.
It's very low for a usually buffered setup, because it uses just a single frame.

[Syntax]

I see that people are attenuating the sync from the strippers they are using, does the GBS not accept TTL sync even though it has a vga input?

[rama]

The 5725 datasheet has the following to say:
Voltage on any input maximum: VDD+0.3 V
That would be 3.6Vpp maximum, already less than a 5V powered LM1881 produces.

Note:
Other chips differentiate between analog and digital pins, with different Vmax for both types.
Seeing how the SOG pin can process a composite video signal, I tend to say that it's an analog input.
So maybe the datasheet is correct. Then the Vmax is 3.6V.
But if they "forgot" to list analog inputs separately, then Vmax is further reduced.

In summary, I believe the amplitude should be well below 1.0Vpp.
I still have to calculate a voltage divider for that and see how it affects the signal slopes and power draw.
Those LM1881 can't deliver a lot of current, so the resistor values need to be a little higher, maybe around 2.2k or so.

[Syntax]

Thanks Rama, I figured that it could take up to 5v considering it has a VGA input plug but is seems all inputs for RGB are linked directly.

Might have to cut the traces for the VGA in and add inline resistors for H V sync.

[rama]

Well, it definitely won't work with 5Vpp signals.
These chips usually accept input signals at or slightly above their Vcc, and the GBS 5725 runs at 3.3V for the most part.

Vertical Sync goes to its own input, which probably is digital. It is terminated with 1kOhm to ground.
Horizontal Sync is shared with RGBS Sync, as you discovered. It also is terminated with 1kOhm to ground.

You need to find out what level your source delivers. You probably need an oscilloscope for that though.
If your source has a series resistor on H / V Sync, it will form a voltage divider with the 1kOhm on the GBS.

[Syntax]

My source will be an Extron mvx so 4-5vpp

[rama]

Yeah, too strong for these boards and probably many others as well.
You may want 1kOhm series resistors in there.

Also, I don't think the software can handle your source. It is separate sync for one (I only once got that working .. a little), and probably has unique timing.

[Syntax]

:/

So putting RGBHV into this is an issue?
Thinking more about it, any system I have that outputs HV goes straight to the vga crt anyway.

I'll be using a gscart to clean sync then try a 1k inline on gbs sync as you suggested.

[rama]

Well, I'm not sure that devices these days output 5Vpp sync signals. The GBS itself for example doesn't: It uses 3.3Vpp instead.

But an old Extron might use 5 volts. It depends on what the standard says and how close they followed it back then.

[rama]

I'm afraid I'm going to have to change the debug-in pin on ESP8266 boards.
Turns out "D5" is connected to an SCK LED, at least on Wemos D1.

The pin will move from "D5" to "D6".
I double checked and this and the other pins in use aren't shared.

Watch GitHub for the commit soon.

[Syntax]

Well, I'm not sure that devices these days output 5Vpp sync signals. The GBS itself for example doesn't: It uses 3.3Vpp instead.

But an old Extron might use 5 volts. It depends on what the standard says and how close they followed it back then.

Vga standard for sync is and always has been TTL 5vpp.

I can use a crappy multimeter to measure it close enough.

Just a suggestion but i think you should delete or archive the first post and sticky the github address there instead, took some hunting to find it the other night.

[rama]

I can't mess with other's posts.
What I could do is open a new thread though. I plan to do this for a stable release, should I ever produce one

For your VGA issue, you may want to check out the schematics of the OSSC. It offers a VGA port, iirc.

[Higgy]

Rama - once the pin switch has been made on Github can you please mention it.

I updated from Github about 9am UK time and the pins were still D5 and D7 (as far as I could see) for Debug and Vsync.

Thanks

[rama]

Just pushed the change.

Sorry for the trouble. I realize that pin changes mean some people have to open up their boxes and resolder wires.
I just didn't notice that this pin was shared.

[rama]

Also recently pushed:
ESP8266 WiFi used to require a hack that changed the MAC address together with the hostname.
This used to be the only way that my router would recognize the changed hostname.
I've changed this to just set the hostname as early in setup() as possible, which fixed the issue with my router.
Hopefully this works for everyone else as well.
If it works, you can access the web ui at http://gbscontrol/ .
If it doesn't work, you need to access it via the assigned IP.

Next I found that the power supply situation is still not solved.
I had my Wemos D1 board powered via the same PSU as the GBS, going to VIN on the ESP8266 (so it uses the onboard voltage regulator) via a diode (so the ESP8266 could not backfeed power to the GBS).
This seemed to work really nice but I just found out that the WiFi module wasn't at all happy.
I was unable to connect to it when it was in AP mode, ie: when it runs the captive portal, waiting to be configured.

For now I'm back to using 3.3V directly from the GBS VCC header (into the 3.3V input on the Wemos D1).
No diodes or anything, so I have to make sure the GBS is always powered on before I connect the Wemos D1 to the pc.
This fixes the WiFi issue and the GBS regulator can handle the load easily.