GBS 8200/8220 CFW Project

The place for all discussion on gaming hardware
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

are there any 3d printed enclosures for the GBS? I wouldn't mind one to complete my setup
fluxcore
Posts: 29
Joined: Fri Aug 01, 2014 10:11 pm

Re: GBS 8200/8220 CFW Project

Post by fluxcore »

AndehX wrote:are there any 3d printed enclosures for the GBS? I wouldn't mind one to complete my setup
https://www.thingiverse.com/search?q=gbs
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

Hey Rama, I've been thinking about other output resolutions that would be nice to have. I've mentioned 1080p previously, for obvious reasons, as it would be nice to have uniform scanlines on my projector. You mentioned it would be quite difficult to add 1080p, but what about resolutions like 960x540 or 720x540? The vertical resolution is exactly half of 1080, which would result in uniform scanlines when upscaled to 1080 by the display right? And you could easily fit a 640x480 scaled image within that. Just food for thought.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Hey,

new output resolutions carry a big maintenance burden right now, as each resolution is its own preset file.
I feel like I want to change this system one day, but it will be a big restructuring effort, with many new things to go wrong.

Once the preset system is generalized though, different resolutions will be easier to maintain.

So I understand the wish to experiment, especially in regards to filters, but I can't commit to this right now.
I have it in mind for the preset system revamp though :)
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

Makes sense. So what's the current plan going forward? What features/fixes are you working on right now?
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Hm, I want to find an SDRAM speed / subtiming mix that has more bandwith and less or the same level of digital noise than before.

SOG (the process of digitizing and separating CSync, either pure or mixed with video) needs to be more reliable.

Odd timing sources (EGA, etc) should have some kind of support, even if the video mode can't be determined.

Component Out is kind of limited, that could be better.

I'd like to find a deterministic phase setting routine. Once that works, the sampling phase should be optimized and corrected automatically.

The entire web ui "experience" is bare bones and confusing. It should have visible feedback about current settings at least.

Well, that's some stuff I think about atm ;p
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

Hello,

Back in August, I had this system working. I did not update it until now because I did use the an old CRT TV instead of the GBS with my old machines do do some demonstrations.
Today, I did update to the latest master commit, even tried the latest commit from dev branch, but I cannot make it working.
No problem to access the web interface, but I can't get a picture more than 10 seconds on my Dell 1908FPb 5/4 screen (1280x1024).

From the COM output, I have messages like
"SD? !HD?"
"lock failed, check debug wire!"
"..........!.!..!...!.!.!.!!!!"

The debug wire is still connected to D6.

I tried to downgrade to an old version from August, it works a bit better but also no more stable like it was months ego, It looses picture after 30 secs.
Tried 2 Amiga : My usual 1200 and a 500 with same result, checked the cable and the SYNC resistor, all are good.

I wonder were the problem can be. The only thing I did change was to put the GBS, power supply and the D1 in a box. The Power supply is 5V 3A.
Image
In the picture, I did disconnect the power pin from the VIN pin because I was powering the D1 via USB to see the console.

"Print Info" with latest version from master: https://pastebin.com/Z6FtnXBw

Edit: And I would like to know if it's possible to connect a SVHS to the GBS (Chroma + Luma) to connect a C64 and if yes, which pin to use?
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Hey,

you can only hook up Luma to the green RCA input. You'll get a black and white picture.
There is no Chroma demodulation in the scaler chip, unfortunately.

"h:315 v:1519 PLL.00 A:171740 S:0.e I:2 D:9da6 m:0 ht:0 vt:0 hpw:000 s:0 W:-72 L:6"
^ This is all broken. Your I2C isn't working.
Do you have a shared ground between the ESP8266 and the GBS?
strayan
Posts: 671
Joined: Sun Mar 19, 2017 8:33 pm

Re: GBS 8200/8220 CFW Project

Post by strayan »

Where do I find a list of features for this project?
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

all the information on this project is here: https://github.com/ramapcsx2/gbs-control/wiki
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

rama wrote:Hey,

you can only hook up Luma to the green RCA input. You'll get a black and white picture.
There is no Chroma demodulation in the scaler chip, unfortunately.

"h:315 v:1519 PLL.00 A:171740 S:0.e I:2 D:9da6 m:0 ht:0 vt:0 hpw:000 s:0 W:-72 L:6"
^ This is all broken. Your I2C isn't working.
Do you have a shared ground between the ESP8266 and the GBS?
Found the problem finally, so stupid one... This board did not have the jumper so I added it but my soldering on the ground pin was not proper and was di-soldered ...
Soldering it back fixed the problem and thanks for the 1280x1024 setting :)

I will now add some filtering caps to try to get ride of moire beginning by the LDO (this board has C11 removed, pots removed, and 100ohm added between sync and gnd).
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

That's a nasty problem to debug. Glad you found it :)

The main input buck converter likes to have more bulk capacitance, but extra bypassing capacitance is most effective if you place it around the scaler chip.
I recommend a few extra 10uF to 22uF SMD caps on the 3.3V supply on all 4 sides of the IC.
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

You mean on C3, C6 and C7 and others?
I have some 10uF SMD ceramic caps ready to be used.

Edit: I did find your old picture, then I did add some 10uF caps to C3, C6, C7 and C42. To be tested :)

Edit 2: Tested and it's a bit better but moire is still largely visible.
My board is this one:
Image

I power the ESP8266 on VIN pin with the same 5V I power the GBS. I have a 22uF capacitor installed between VIN and GND.
Do you have other suggestions? I have only big 220uF tantalum capacitors, so I cannot use them for the PDO I think.
Last edited by Rodrik on Wed Jan 02, 2019 2:41 pm, edited 1 time in total.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

I'm not sure that the capacitor numbering is the same on all boards, but on the board I have out now,
I'd place the 10uF besides C23, C28, C41, C48
C28 is the 1.8V digital supply, the others are 3.3V around the scaler chip.
Especially important would be C23, where the single 3.3V supply runs by.
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

Here are photo of the not so good soldering, but no short-circuit.
Image

Image

So C23 and C3 seams to be on same traces, so this one is done.
Same for C28 and C7.

And on another subject, using the latest master version, I have this small issue at the right of the screen all the hight long: One pixel down shift.

Top
Image

Bottom:
Image
Last edited by Rodrik on Wed Jan 02, 2019 3:23 pm, edited 1 time in total.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Well, not all ideal placements but overall it's going to be effective :)

These extra caps will reduce SDRAM switching noise but there are many sources of interference in the wild.
An easy thing to try out is changing the memory clock. This is an option on the bottom of the web ui.
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

I could not find better picture than "feedback clock" in memory clock.
I have 10uF SMD ceramic, 220nF SMD ceramic, SMD ferrite beads and tons of regular electrolytic caps at my disposition if you have other idea.
Higgy
Posts: 54
Joined: Thu Mar 15, 2018 1:29 pm

Re: GBS 8200/8220 CFW Project

Post by Higgy »

Rodrik - I've just received another GBS board to Mod from scratch and its the same model v5.1 as yours. Just waiting for some more parts before I can start modifying. Are you using a Sync stripper or have you replaced Caps C33 & C35?

I was using my older GBS last night on my AMIGA with Master GBS Control software (currently code is 25 days old). I did not notice any shift. I will try and test again tonight as it is still setup on my table.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Okay, but the noise pattern did change with the memory clock settings?
If so, the switching noise is still strong enough to couple into the ADC inputs.

I don't know about all the details of your setup, so here's some info you may be able to use:
Image
Image

The big plane below the chip is all 3.3V. Most of the area on the underside is a ground plane.
The design isn't bad at all, but there is this flaw that the 3.3V service both, analog and digital.
Your bypass capacitors are mostly needed on the north and east side (towards the SDRAM), where the digital switching happens.
The south is the analog area.
You want that to be noise free (so for example, avoid placing copper foil over the digital area that overlaps into the analog area).

For the analog side, you want to make sure your RGB cable has good ground returns and is generally well constructed.
No pin headers with long, single strands of wire if possible.

With regards to the shifted pixels, that source exhausted the glitch free active video zone.
It simply produces too many pixels, possibly from stretching horizontally?
Also, this will happen if you move the picture too much. There is very little margin in some presets.

Edit:
By the way, this is my first GBS from 2013 or so? It has seen things.. :D
Higgy
Posts: 54
Joined: Thu Mar 15, 2018 1:29 pm

Re: GBS 8200/8220 CFW Project

Post by Higgy »

Just tested my AMIGA 500 on Workbench 1.3 screen and I see no pixel shift.
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

Higgy wrote:Rodrik - I've just received another GBS board to Mod from scratch and its the same model v5.1 as yours. Just waiting for some more parts before I can start modifying. Are you using a Sync stripper or have you replaced Caps C33 & C35?

I was using my older GBS last night on my AMIGA with Master GBS Control software (currently code is 25 days old). I did not notice any shift. I will try and test again tonight as it is still setup on my table.
No Sync stripper and no change for Caps C33 & C35 because I miss the parts. C33&35 are planned if I find the 1nF caps, but I did not think this could alter the picture. Is it?
About my setup, you can see the picture upper on this page. I use the wires provided with the GBS soldered to a DB23 used by my Amiga 1200, 500 or CD32 (with terrible fire addon) and connected to the PIN header behind the input VGA port of the GBS, should I use another solution?

I did not connect anything else than the Amiga to the GBS yet because I used the provided wires to build an Amiga cable, I will make a Scart input to test for example my TO7/70 in a week or 2.

@rama, no the noise pattern does not change with memory clock, with very low clock the picture is greenish with vertical green lines. With high clock the noise is lot more visible.
It's many vertical S noise, less visible than the old GBS I have with dual output but still very visible. The more black the display is, the more the noise is visible. No noise visible on white pictures.
For shifted pixels , I will reduce the width stretch and ask the monitor to calibrate itself to fill the screen.

I did have some added shielding on the old GBS, but I did remove it strait after you told me to do so back in July; and I have nothing like this on the V5 one, I do only the mods you suggest.

Many thanks for your help, I am sorry if I did upset you with all my questions. Going back to work tomorrow, so I will not bother you again until Saturday :lol:
Last edited by Rodrik on Wed Jan 02, 2019 8:10 pm, edited 2 times in total.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

It's all good, you guys provide valuable feedback ;)

I just merged the development branch to master.
Memory subtimings changed and due to the random elements with noise (different on each board), maybe the new timings work better.

The noise could have developed somewhere earlier in the chain. If you have other cables, try those.
Or try other machines and see if they look any better.

The web ui has an actual theme now, hope you guys can cope with it ;p
Rodrik
Posts: 18
Joined: Tue Jul 03, 2018 12:28 pm

Re: GBS 8200/8220 CFW Project

Post by Rodrik »

Thanks Rama, will update on Saturday.

And you know what? I fixed the shifted pixels. I did still have overscan settings in the Workbench settings, I did remove it and set it back to 660x268 and it fixed the problem after I did shift the picture to right using the webui and back to left using Workbench overscan setting.
rama
Posts: 1373
Joined: Wed Mar 08, 2017 3:15 pm

Re: GBS 8200/8220 CFW Project

Post by rama »

Cheers ;)

I went back and probed the limitations of active video (with regards to the overscan problems).
There were a few more pixels in the 1280x presets to be had, so they're now available.
From that Amiga Workbench picture, it may be almost enough to fix the picture shift.
It looks like this is the limit now though, at least when keeping the no compromises approach on quality.
(It's still possible to find good looking results with a reduced sampling clock or non-ideal horizontal scaling.)

Also, the besthtotal routine that adjusts the presets for different sources is now smarter.
It used to accumulate rounding errors when switching back and forth between 2 states, such as a source switching between 240p and 480i.
This is working well now, which helps with frame locking and especially towards more unified presets.
To do this, I had to redo the horizontal timing calculations a little.
It should be better than before, but I couldn't test everything and there may be bugs still ;p
User avatar
NoAffinity
Posts: 1018
Joined: Mon May 07, 2018 5:27 pm
Location: Escondido, CA, USA

Re: GBS 8200/8220 CFW Project

Post by NoAffinity »

rama wrote:Cheers ;)

I went back and probed the limitations of active video (with regards to the overscan problems).
There were a few more pixels in the 1280x presets to be had, so they're now available.
From that Amiga Workbench picture, it may be almost enough to fix the picture shift.
It looks like this is the limit now though, at least when keeping the no compromises approach on quality.
(It's still possible to find good looking results with a reduced sampling clock or non-ideal horizontal scaling.)

Also, the besthtotal routine that adjusts the presets for different sources is now smarter.
It used to accumulate rounding errors when switching back and forth between 2 states, such as a source switching between 240p and 480i.
This is working well now, which helps with frame locking and especially towards more unified presets.
To do this, I had to redo the horizontal timing calculations a little.
It should be better than before, but I couldn't test everything and there may be bugs still ;p
These both sound like great improvements! If I get a chance this weekend, I will test with a 240p/480i title. My capture hardware (elgato camlink) doesn't always like it even when the the signal change is on the other side of a scaler - that little blip while the scaler re-locks is enough to throw it off. Oddly, those cheap chineses SCART-to-HDMI scalers do a pretty good job of "smoothing" the signal and keeping the output consistent through the transitions. Will be interesting to see if GBS CFW, with all of it's benefits, can now also rival this feature. :)
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

NoAffinity wrote:If I get a chance this weekend, I will test with a 240p/480i title.
Resident Evil 2 on the N64 with the expansion pak is a good one to test with.
User avatar
NoAffinity
Posts: 1018
Joined: Mon May 07, 2018 5:27 pm
Location: Escondido, CA, USA

Re: GBS 8200/8220 CFW Project

Post by NoAffinity »

AndehX wrote:
NoAffinity wrote:If I get a chance this weekend, I will test with a 240p/480i title.
Resident Evil 2 on the N64 with the expansion pak is a good one to test with.
Will this work in attract mode? I'm looking to be as lazy as possible with my testing, and throw something on attract mode while recording, and while doing other things. 8)
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

hmm possiby? I dunno, try it :P
Higgy
Posts: 54
Joined: Thu Mar 15, 2018 1:29 pm

Re: GBS 8200/8220 CFW Project

Post by Higgy »

From Eurogamer's 'Digital Foundry' video on Resident Evil the N64 version changed a lot. You should see the change going from the street to the gun shop.

Anyone else updated to today's CFW? My Arduino software picked up 2 Updates regarding the Wi-Fi and I got compiling errors. Ran out of time for further testing. Probably needs all flash programming.
User avatar
AndehX
Posts: 790
Joined: Sun Oct 18, 2015 11:37 pm

Re: GBS 8200/8220 CFW Project

Post by AndehX »

I can flash it now and give it a try
Post Reply