GBS 8200/8220 CFW Project

[rama]

Ok.
I'd expect an Ethernet enabled module to come with an Arduino library that would have to be included.
The library would probably require a bit of configuration, but then it all should work the same as with WiFi.
I'd say 2-3 development hours, and the system should work

The bigger issue is with finding a suitable Ethernet module.
Ebay has some "Wiznet W5500" modules that connect via SPI bus.
SPI needs 3 additional pins on the ESP8266, and there is a reset and a CS signal that may be required.

Whether Ethernet for gbscontrol will be possible comes down to pin use and some creativity.

[AndehX]

ok so the newest gbscontol seems more stable with the sync stripper on my gscart disabled, with it enabled, it's still glitchy. I've noticed the 256x239 mode on the SNES 240p test suite still produces the glitch, even with the sync stripper disabled.

[rama]

SNES 239 mode and the Mega Drive 256/320 modes seem to be the 2 extremes that a sync processor has to deal with for SD formats.
For SNES, I bet the 239 mode exposes the known jitter scanline issue more.
For MD, its 2 modes have a large HSync pulse width difference that needs attention when setting coast/sync pulse ignore windows.

I know that the gscart can do microprocessor controlled sync regeneration or bypass CSync.
What I don't know is whether that bypassed signal is from the sync stripper, or actually from the source console.
It's entirely possible that the sync stripper itself can't deal with SNES or MD anomalies well.

Ah well, sorry for the rant! :p
I need to look at the gscart to tell for sure.

[rama]

Just an update:
I've reworked automatic SOG level adjustments and tight H-coast windowing.
This is now super resiliant against bad sync pulses, noise and low level signals.
In my tests, it even once managed to lock on to the HBlank period of a signal without any true sync pulse!
(This would happen if you just connected RGB (but no "S") to the RCA inputs.)

It needs lots of testing, but then it should be able to sync to everything!
Well, everything within reason ;p

Edit:
Oh, and since I just ran into this weird "bad sync" behaviour with 1080p: Check your power supply and cabling!
A weak link in the barrel connector can easily look like a sync problem. I just had to rotate the plug a few times and the issue was fixed.
The power input barrel connector is simply not of the best quality..

[NoAffinity]

Thank you for all the continued improvemnets rama!

Gonna update and run it through some paces. I have a weird thing going on with my model 1 hdg genesis - after its.been on for more than a couple seconds, if i power cycle, ossc and older build of gbsc (a couple weeks old) wont see a signal. Arcade otaku scart switcher does indicate a signal is passing through tho. Hoping this new gbsc update will give some.improvement in this area.

On another note, any recommendations for a good repepacement for the barrel connector? Dont think ive ever had a problem with mine but i do like to improve things.

Sent from my SM-G955U using Tapatalk

[rama]

Hm, no. It might just be an issue with my cable, too.
The inner diameter of these barrel connectors has to match the male end perfectly, and if it's a little off, it's hard to realize that.
All I can say for sure is that while twisting the plug a little, power is intermittent.

[rememberizer]

rememberizer:
Well, the update is already out.
This tells me that the video mode isn't "unknown", but that the sync isn't within parameters to be considered stable.
Could you share your RPI setup and the text file that sets the video mode?
I have a Pi2 and Pi3 here.

Oh okay, I'll try it again right now. I haven't tried since the last time!

Using my Raspberry Pi 4, I'm getting this in the box and a messed up screen from my capture card:

Code: Select all

Source format not properly recognized, using fallback preset!
ADC offset: R:40 G:40 B:40
post preset done (preset id: 3) for 60Hz EDTV

This is the timing line I'd like to use in config.txt:

Code: Select all

hdmi_timings=1920 1 152 246 280 240 1 3 7 12 0 0 0 60 0 40860000 1 #SUPERRESO

I'll try the Pi 3 in a bit! I got exactly the same result with the Pi3 using a CRT Retropie image.

For some reason, it's completely unable to sync with CPS2 now. I used to be able to do so easily on the same setup. CPS3 and Namco System 12 still work! Edit: this only happens when I'm on the arcade cab capture setup with a passive breakout, and it turns out it still happens with the previous version of gbscontrol. Funny because it did work when I tested it before.
Edit2: Nevermind, it worked when I plugged the sound in!

[rama]

hdmi_timings=1920 1 152 246 280 240 1 3 7 12 0 0 0 60 0 40860000

^ Could you Google decode this for me, please?
I think it says the resolution should be "1920x240", which is fine, but there is this "280" in it that I can't explain.
If the "280" is meant to be the physical vertical line count, then this is wrong and should be around 263.

Edit2: Nevermind, it worked when I plugged the sound in!

Very suspicious..
Make sure all your grounding is good.

[DarkAries]

I did finally get my GBS-8220 setup with the CFW. Works great! I did try out a few things, like trying to scale PSP to fill the screen. After a certain amount of sizing on the horizontal you just end up getting garbage. Overall loving the device.

[rama]

Glad you like it, DarkAries.
The scaling and repositioning options have some limits, as you've noticed.
It's usually good enough to make drastic changes but size doubling or anything like that is too much.
It's better to try a different output resolution first. See if any of them produces better results.

[rememberizer]

hdmi_timings=1920 1 152 246 280 240 1 3 7 12 0 0 0 60 0 40860000

^ Could you Google decode this for me, please?
I think it says the resolution should be "1920x240", which is fine, but there is this "280" in it that I can't explain.
If the "280" is meant to be the physical vertical line count, then this is wrong and should be around 263.


Very suspicious..
Make sure all your grounding is good.

The line is just taken from guides I found online. I have no idea what that even means! What do you suggest I use? I just need the pi to work on the cab and the scaler.

Edit:
From the raspberry pi site:

Code: Select all

hdmi_timings=<h_active_pixels> <h_sync_polarity> <h_front_porch> <h_sync_pulse> <h_back_porch> <v_active_lines> <v_sync_polarity> <v_front_porch> <v_sync_pulse> <v_back_porch> <v_sync_offset_a> <v_sync_offset_b> <pixel_rep> <frame_rate> <interlaced> <pixel_freq> <aspect_ratio>
     

https://www.raspberrypi.org/documentati ... t/video.md


Yeah the grounding is a bit confusing since I'm using different devices for sound and video capture.

[rama]

That's what I needed, and the numbers all seem correct.

I couldn't find a definitive specification of the sync polarity bits though (your timings have HSync and VSync polarity as "1").
Try the following line just for a quick test, please:

Code: Select all

hdmi_timings=1920 0 152 246 280 240 0 3 7 12 0 0 0 60 0 40860000 1 #SUPERRESO

[Iraito]

Is it possible to use composite with the GBS ? i know it's not ideal but for NES\famicom stuff i prefer it.

[rama]

Nope, that requires an external CVBS decoder.

[Iraito]

Well good to know at least, i guess i will go with one of the cheap upscalers only for composite.

[rama]

Honestly, you will be disappointed by any of those.
Lag is through the roof and then they don't even look all that great.

Get this one for decent results
https://www.videogameperfection.com/pro ... rotink-2x/

[rememberizer]

That's what I needed, and the numbers all seem correct.

I couldn't find a definitive specification of the sync polarity bits though (your timings have HSync and VSync polarity as "1").
Try the following line just for a quick test, please:

Code: Select all

hdmi_timings=1920 0 152 246 280 240 0 3 7 12 0 0 0 60 0 40860000 1 #SUPERRESO

I tried this and my cab couldn't sync. The image was rolling vertically. The GBS Web UI was giving me "no signal" messages this time. Let me know if you want me to test anything else!

[NoAffinity]

Ive tested with dreamcast, snes (sns-101, rgb modded, dejitter modded) and geneses (model 1 hdg, va6, bypass modded). All working and looking great.

Did have some weird bouncing on the genesis. Testing contra hard corps attract mode, when it would cycle back to the title screen it would do it. If i cycled my scart switch to another input and back, it would work fine. That was sunday. Unable to replicate it tonight.

Sent from my SM-G955U using Tapatalk

[rama]

It's always the Genesis, model 1 specifically.
I wonder what the problem with this console's CVBS Sync is, specifically.
My RGB cable has a straight through CVBS sync line, and when looking at the signal under the scope, it tends to look horrible.
This is also the console with the widest HSync width swing when the picture is white vs black.
You can easily see the picture wander quite a few pixels when the image content changes.

Maybe the problem is that CXA1145...

[Iraito]

Honestly, you will be disappointed by any of those.
Lag is through the roof and then they don't even look all that great.

Get this one for decent results
https://www.videogameperfection.com/pro ... rotink-2x/

I will take that into consideration, for now i will either go CRT or directly on my LCD TV in gaming mode, the retrotink definitely looks like the best solution for composite, but of all my consoles the famicom is the only one that requires it, i don't feel like buying it now, especially since i had to buy a famicom + 72 to 60 pin adapter + flashcart.

[DarkAries]

Is there a way to completely start from scratch again? I was playing around with the 960i output from the gamecube via swiss, hit too many buttons and now can't even get an image on the default 480i output of the console. Tried reuploading the sketch but it seems all settings persist.

Also 960i output from the cube just results in a diagonal rolling stuttering image. Sad times.

Edit: 240p sources like a genesis work fine. But any outputs from the cube are apparently unrecognized now and I can't get anything but garbage video now. The video works fine direct to the tv and everything so.

[NoAffinity]

It's always the Genesis, model 1 specifically.
I wonder what the problem with this console's CVBS Sync is, specifically.
My RGB cable has a straight through CVBS sync line, and when looking at the signal under the scope, it tends to look horrible.
This is also the console with the widest HSync width swing when the picture is white vs black.
You can easily see the picture wander quite a few pixels when the image content changes.

Maybe the problem is that CXA1145...

See my setup here: https://r.tapatalk.com/shareLink/topic? ... source=app. The cxa1145 is removed. Bypass modded from the sega video chip, with all pins lifted off the pcb.

I was able to replicate it, but again switching to a different input at the scart switch and back resolved it. Only on contra hard corps so far, and when it transitions to the title screen. Not a big deal just interesting.

Sent from my SM-G955U using Tapatalk

[rama]

DarkAries:
You're looking for the reset defaults button (web ui > system tab).
I'm not sure 960i is a valid format for SCART RGB, btw.

NoAffinity:
Check what kind of buffering (if any) your CSync output has.
Is that a 220uF capacitor and a 470 Ohm series resistor in the cable?
I can check the game here with my machines, but I think the problem is with the CSync setup somehow.
Make sure it's all correct, so that at least an undisturbed 0.3Vpp (or more) signal enters the scaler chip.

One more thing:
Please post a few lines of information mode for when it works and for when it doesn't.
I'm interested in the "hpw" variable.

[AndehX]

Also 960i output from the cube just results in a diagonal rolling stuttering image. Sad times.

960i is only usable with an HDMI mod. It will not work over SCART.

[NoAffinity]

NoAffinity:
Check what kind of buffering (if any) your CSync output has.
Is that a 220uF capacitor and a 470 Ohm series resistor in the cable?
I can check the game here with my machines, but I think the problem is with the CSync setup somehow.
Make sure it's all correct, so that at least an undisturbed 0.3Vpp (or more) signal enters the scaler chip.

One more thing:
Please post a few lines of information mode for when it works and for when it doesn't.
I'm interested in the "hpw" variable.

Yes, it's a 220uf cap and 470 ohm series resistor on sync.

Here's a capture of the issue: https://www.youtube.com/watch?v=_O6fFbk ... e=youtu.be

Will get you some info mode lines.

[rama]

That's some very unusual bouncing. I haven't come across this behaviour yet.
With your cable configuration, your CSync definitely has to be buffered.
The buffer is not on your RGB bypass board, since that board uses a 3 channel amplifier, and the colors already occupy all channels.

So it'd be best if you found your old installation guide and find out how CSync is handled with this mod.
I know for "pretty sure"(;p) that the stock CSync output won't be good enough to drive this cable.
(This is a problem with all stock Mega Drive / Genesis CSync outputs. They're not designed for direct use, but the retro community does it anyway. The margin for error is very small.)

[DarkAries]

I should have mentioned, my bad. I use the component input. I haven't even touched the RGB input. I tried the reset but it still can't detect any output from the gamecube. I'm thinking I potentially blew something up. I did a complete reflash and even an older build but nothing. When I get the chance I'll try with a wii, ps2 or something. The dev console shoes that it detects activity on the input but there is never a signal it can actually do something with it seems.

Edit: Okay so it seems after messing around with more things, the signal is there, it looks like its just not being detected properly. When the console reports a signal it has a very low vertical size. And any output I can coax out of the thing looks like such. Its so weird since it was working fine earlier, and then I started to just see what video modes the GBS supports and now it won't lock on at all. This is just regular 480i from the gamecube on the menu.

Code: Select all

h: 511 v:  28 PLL:00 A:7b7b7b S:02.18.00      I:85 D:0000 m:0 ht:   0 vt: 113 hpw:2349 u: 96 s: 0 S:13 W:-68

[rama]

There is a SOG level control button on the web ui development tab. When you hit it, the "S:" variable goes down from 13 to 0 in steps of 1.
Try going to level 2 or 1 and see if it improves.

[NoAffinity]

That's some very unusual bouncing. I haven't come across this behaviour yet.
With your cable configuration, your CSync definitely has to be buffered.
The buffer is not on your RGB bypass board, since that board uses a 3 channel amplifier, and the colors already occupy all channels.

So it'd be best if you found your old installation guide and find out how CSync is handled with this mod.
I know for "pretty sure"(;p) that the stock CSync output won't be good enough to drive this cable.
(This is a problem with all stock Mega Drive / Genesis CSync outputs. They're not designed for direct use, but the retro community does it anyway. The margin for error is very small.)

Well, sadly, I got this mod off ebay quite some time ago, and when I went back to the history in my ebay, I'm unable to get to any useful info. Not sure who made the mod, other than a seller named nausicaaandme.

Anyway, not sure if this is helpful, but here are the info mode reports.

While it is bouncing:

Spoiler

h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 267 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0594 m:1 ht:2345 vt: 257 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 268 hpw: 215 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 264 hpw: 215 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0500 m:1 ht:2345 vt: 268 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 262 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 263 hpw: 214 u: 0 s:13 W:-65
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 258 hpw: 214 u: 0 s:13 W:-65
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 262 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 264 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 273 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 265 hpw: 214 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 265 hpw: 214 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 264 hpw: 214 u: 0 s:13 W:-65
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0500 m:1 ht:2345 vt: 260 hpw: 214 u: 0 s:13 W:-65
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 266 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:058c m:1 ht:2345 vt: 264 hpw: 214 u: 0 s:13 W:-64

After cycling scart switch inputs, and bouncing has gone away:

Spoiler

h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-62
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 214 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-62
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 214 u: 0 s:13 W:-62
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 214 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0500 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 214 u: 0 s:13 W:-63
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 214 u: 0 s:13 W:-64
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:058c m:1 ht:2345 vt: 259 hpw: 214 u: 0 s:13 W:-62
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0500 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-62
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 214 u: 0 s:13 W:-62
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-61
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 215 u: 0 s:13 W:-61
h: 429 v: 523 PLL:00 A:7b7b7b S:8f.10.00 H- I:00 D:0584 m:1 ht:2345 vt: 259 hpw: 214 u: 0 s:13 W:-61

And, the only time I've seen this is with Contra Hard Corps, and when it cycles back to the beginning of attract mode, wherein the characters "slide" onto the screen:

Spoiler

[rama]

Okay, thanks.
These logs seem like the SOG slicer is just fine.

The problem is probably with the coast window, VSync detect window or pulse ignore length.
These are values that I need to tweak so they work with all possible sources.

Can you arrange a Teamviewer session with the ESP8266 connected via USB and controllable via Arduino IDE?