Sophia 2 - improved GTIA replacement

[Johny_lovin_it]

On 7/15/2022 at 4:47 AM, Beeblebrox said:

@retrofez Go and log interest on the following thread, then wait till 2024 as the global chip shortage is dictating the next batch runs it's creator Simius can build. 2024 is the predicted date when he'll be able to get his hands on the chips. There is a massive preorder list.

 

 

I guess I'll be on the preorder list as well. I have an Atari XEGS and I'd like to have it rgb modded too.

[+Philsan]

On 7/10/2021 at 5:55 PM, mdivancic said:

Installed the longer cable that @orpheuswaking made for me and buttoned up the 800xl. Using the 3D printed part designed by MrLurch on Thingiverse makes for a nice clean install. Pardon the messy work area.

 

 

Could you please share the link? I don't find the file on Thingiverse.

[fenrock]

39 minutes ago, Philsan said:

Could you please share the link? I don't find the file on Thingiverse.

Do you mean this one? https://www.thingiverse.com/thing:4869472

I just searched "DVI Atari" and it was one of the 4 hits.

[Beeblebrox]

That's the one I have I installed in my other 600XL. Very robust. Wiggles a little in the pbi slot in the casing. It works well although I had to extend my ribbon cable as was a little too much of a stretch. It just about stretches but I added a 10cm section of ribbon cable into the mix and now it isn't taut and is a little better routed inside. 

 

 

[+Philsan]

5 hours ago, fenrock said:

Do you mean this one? https://www.thingiverse.com/thing:4869472

I just searched "DVI Atari" and it was one of the 4 hits.

Thanks! I searched with "Sophia 2".

[fenrock]

1 hour ago, Philsan said:

Thanks! I searched with "Sophia 2".

Fear my search-foo

[phonedork]

I just got one to replace my original Sophia DVI. I like how this one uses a rom to change resolutions. Will this also support RGB 240p /15khz? If so will this just require a VGA adapter?

[phonedork]

There is not a whole lot of info about the Sophia 2 install. Luckily I had a friend of mine install the previous Sophia board and socketed the chip. So I just replaced it with the Sophia 2 board and 24+5 dvi. The 24 +5 dvi is the main difference I believe since the +5 can do analogue. I’ll find out when o get my dvi 24+5 to vga. I’m hoping to get rgb 15kz. The other thing is the Sophia 2 rom. Pressing 4 you can select between options. But the * and # buttons cycle too quickly sometimes to choose an option like different resolutions. 
 

anyways here are some pics 

[+-^CrossBow^-]

18 hours ago, phonedork said:

There is not a whole lot of info about the Sophia 2 install. Luckily I had a friend of mine install the previous Sophia board and socketed the chip. So I just replaced it with the Sophia 2 board and 24+5 dvi. The 24 +5 dvi is the main difference I believe since the +5 can do analogue. I’ll find out when o get my dvi 24+5 to vga. I’m hoping to get rgb 15kz. The other thing is the Sophia 2 rom. Pressing 4 you can select between options. But the * and # buttons cycle too quickly sometimes to choose an option like different resolutions.

As I had advised on FB (I believe you are the same person...) there is a newer updated version of the configuration program for the 5200. It is in the second to last page of the initial Sophia 2 pre-order thread.
 

 

sc5200.rom

[TheMontezuma]

On 12/1/2020 at 8:20 PM, Simius said:

How to mount the Sophia 2 on the Atari 800

 

5. Drill two 1mm holes next (5mm) to the old + holes for radial 2200uF capacitors and mount them with two additional wires.

 

 

 

 

 

 

 

I just went through the SOPHIA2 installation in the A800 NTSC computer. Everything done according to instructions. Everything seemed to work perfectly.

Then I inserted PokeyMax and realized that SIO communication does not work (strange sound/noise, RespeQt shows a lot of issues).

 

To make the long story short:

 

Modifications of the A800 Power Board (STEP 5 - see above) cause issues with SIO DATA lines!

 

The original POKEY chip seems to be robust enough to handle the interference on DATA IN while ATARI is sending data on DATA OUT.

However PokeyMAX does not. SIO communications fails.

 

Guilty is the capacitor C204, which is very close to the SIO lines:

 

I moved the cap connectors to the top side the Power Board and moved the capacitor itself to the middle, so the + and - wires have approximately the same length.

And most important: I took aluminium foil, folded it in several layers and inserted it between the capacitor and the board, creating a shield isolating SIO lines from interferences.

 

 

The shield helped - I could load software from SIO2SD even with Pokey divisor 0.

The oscilloscope shows only minor interferences (I saw the same on my other ATARI computers I tested):

 

 

I hope that my frustrating experience will help other users.

 

 

 

[flashjazzcat]

Good fix. Had to remove the FCC-mandated SIO caps on a 600XL which would do divisor 0 perfectly well with the caps in place and a real POKEY chip, but not with PokeyMAX until the caps were removed.

[Faicuai]

Ok, after flawless and pleasing use of Sophia-2 on my 800/Incognito, I've just found something really odd.

 

So before I raise the BUG-flag here, let's triple check here:

1. Can anyone please, try launching CRUSH-CRUMBLE-and-CHUMP! (?) For this purpose, please use:

• Either Basic rev.C, Altirra Basic v1.57/1.58, Turbo Basic, etc.
• Launch from ATR image using OS/b
• Or mount as D1, launch SDX and manually load .BAS right from Basic-car. session.

1. Do you SEE the gold/yellow Godzilla sprite on the splash-screen and later on the preview-sequence?
2. Here are the images you should try (whatever is easier):

• Crush Crumble and Chomp.atr
• Crush, Crumble and Chomp!.cas

On my end, the digital video-output from the systems CRASHES, and when it does so, all Sophia settings in BIOS are reset to 480p/576p, 16-lumas disabled, etc. The yellow Godzilla sprite NEVER shows up, on either splash-screen or preview-sequence.

 

Thanks to anyone for taking the time to test this...

Edited May 17 by Faicuai

[Faicuai]

Let's see if anyone else has had the chance to test this first, before I raise the "bug-alert" ... on Sophia-2...

[Simius]

It isn't a Sophia's bug but writing #$FF to GRACTL

 

B25A   LDA D4      (FF)

B25C   LDY #00

B25E   STA (95),Y  (D01D+Y)

B260   RTS

 

 

[Faicuai]

37 minutes ago, Simius said:

It isn't a Sophia's bug but writing #$FF to GRACTL

 

B25A   LDA D4      (FF)

B25C   LDY #00

B25E   STA (95),Y  (D01D+Y)

B260   RTS

 

 

So could you replicate the issue, as well?

 

Mind you that it WORKS on my reference 800 (1980-0077) running on standard GTIA.

 

So the only question is (pardon my limited background on Sophia I/O dependencies) why does it work with standard GTIA and why not with Sophia-2?

[Simius]

See the datasheet. 

 

[Faicuai]

2 hours ago, Simius said:

See the datasheet. 

 

This is great, THANKS !!!

 

Would this be cured by having C-C-Chump writing #00000111 (or $07) instead?

 

The thing is that $A000-$BFFF is occupied by Basic rom, during CCChomp runtime, so your trace cannot be pointing to the game's code, it seems... 😳😳

Edited May 20 by Faicuai

[Simius]

It's a BASIC ROM, specifically the POKE command:

 

JSR LETNUM

LDA FR0

STA POKADR

LDA FR0+1

STA POKADR+1

JSR GETBYT

LDA FR0

LDY #$00

STA (POKADR),Y

RTS

 

This means the game's code simply contains POKE GRACTL,$FF

 

INIT:

80 POKE 54279,116:POKE 704,46:POKE 705,46:POKE 53256,0:POKE 53257,0:POKE 559,46:POKE 53277,255

 

CCC65:

1800 GRAPHICS U0:POKE 54279,116:POKE 559,46:POKE 53277,255:POKE 756,120:GOTO 5000

 

 

 

 

 

 

Crush Crumble and Chomp.atr

[_The Doctor__]

so anything setting GraCTL to $FF will set Sophia down the 'special feature' garden path?

[Rybags]

Anything setting the top bit actually.

But really, there's no reason legacy software should put invalid values in GRACTL.  Generally the register is a set once usage case, mainly when you're setting up to use PMGs.

The exception being the TRIG latch (bit 2) where you would toggle it off to reset the latching - though I can't think of any software that uses the TRIG latching and there's no point using it for games.

[flashjazzcat]

49 minutes ago, Rybags said:

But really, there's no reason legacy software should put invalid values in GRACTL.

Exactly, and hence the unlock mechanism works as intended 999 out of 1,000 times. Much like the repeated register 'bug' in U1MB, which was also pretty much a non-issue until parties intent on making a song and dance about non-standard hardware behaviour and relentlessly exploiting same came along.

[Faicuai]

11 hours ago, Simius said:

It's a BASIC ROM, specifically the POKE command:

 

JSR LETNUM

LDA FR0

STA POKADR

LDA FR0+1

STA POKADR+1

JSR GETBYT

LDA FR0

LDY #$00

STA (POKADR),Y

RTS

 

This means the game's code simply contains POKE GRACTL,$FF

 

INIT:

80 POKE 54279,116:POKE 704,46:POKE 705,46:POKE 53256,0:POKE 53257,0:POKE 559,46:POKE 53277,255

 

CCC65:

1800 GRAPHICS U0:POKE 54279,116:POKE 559,46:POKE 53277,255:POKE 756,120:GOTO 5000

 

 

 

 

 

 

Crush Crumble and Chomp.atr 90.02 kB · 1 download

THANKS!!

 

Did not download the image, just went straight to INIT and CCC65 Basic files and updated lines 80 and 1800, respective ($07 on GRACTL works perfectly !!).

 

Thankful for your time tracing this. I really need to get my act around current debugging tools, as I should have (at least) being able to trace down the specific places where GRACTL was being set on CCChump. 

 

This case may also be a reminder of how to even further harden (in the future) access to SPECEN (and features) in Sophia-2. It was surprisingly simple for a rogue $FF to alter course of things, unexpectedly. In retrospect, writing a careless $FF to GRACTL back then may have been "Ok", since GTIA bits 7-3 were presumed "dead",  by specs.

 

Any other piece of legacy SW doing this, will crash as well. (I have NOT found it, though).

 

Cheers! 

Edited May 21 by Faicuai

[Simius]

I could make the access more complicated, but I didn't want because I wanted to make switching color palettes as fast as possible.