Raven. A homemade Atari-like computer

[Oldskool]

Thanks for the files.
My friend has a microscope and all the fancy soldering stuff.
He can look at those IC's and do a reflow.
Hereby the output of the two new roms.
Still no pictures some errors:

dbg2.txt

(3.69 KiB) Downloaded 44 times

dbg3.txt

(3.69 KiB) Downloaded 36 times

[agranlund]

What would be ideal is an ISA card which has a GEM VDI driver on-board in ROM and exposes dual-ported RAM, with one area overlaying the ST graphics addresses in bus snooping mode for compatibility.

Yeah, they all pretty much already have rudimentary drivers on-board but they are of course for PC.
At least all the Init stuff and for very easily putting the card into any of the standardised PC display modes using bios interrupt API.

Pretty good stuff, shame we can't use it (unless we run the cards VGA-Bios through an x86 emulator - and I think that's how we actually init PCI cards on Hades, CTPCI and so on. And even on modern x86 computers in some situations)

On PC, the Mach32 can be toggled between VGA and 8514 mode with the VGA-Bios exposed API because of it.. no need to know the ins and outs of the cards registers.. I'd love to have that

[stephen_usher]

Oh, I forgot that the ISA bus is only 8 bit.

No it's 16bit, but yeah that's still only half of what you'd ideally want

Oh, in that case snooping the ST RAM video memory address region for a boot-time frame buffer would work. (Shadow Shifter?)

For GEM it just needs VDI commands/calls, via a shim driver layer, to be implemented directly by the FPGA/PiPico/Whatever implementing DMA to read any data from main memory but doing all the hard work and blitting etc. internally and outputting to its native HDMI.

Sorry, just thinking out loud.

[agranlund]

Hereby the output of the two new roms.

Good stuff!
Looks like you have something fishy going on with bit6 on the databus. The readback after write verification indicates that this bit is always being read back as a zero.

Code: Select all

Readback Written
 v        v
10110100 11110100       b4!=f4
00011111 01011111       1f!=5f
00001111 01001111       0f!=4f
00010000 01010000       10!=50
00010100 01010100       14!=54
00000000 01000000       00!=40
10101010 11101010       aa!=ea
10011111 11011111       9f!=df
10100111 11100111       a7!=e7
10000011 11000011       83!=c3
10111111 11111111       bf!=ff
00010011 01010011       13!=53
 ^        ^

I was going to ask you to run the normal rom, and press the NMI button after we think it has arrived to the desktop. And then execute commands to write + readback from gfxcard RAM, but with what we have in the log I don't think that is necessary.

My strongest suspect based on that would that either BD30, BD22, BD14, BD6 on U103/U106. They are all neighbours to a GND pin and if shorted to it would result in what we see in the log.
I would check the corresponding data pins on the CPU-facing side of the buffers too, but they feel less likely since your have ROM+RAM working, but who knows.

Screenshot 2024-08-02 at 17.08.37.png (87.62 KiB) Viewed 860 times

The issue could be on the peripheral side of the 68150 bus too but it's a lot harder to short anything there compared to on those fine-pitched buffers.

Edit: omg I just noticed the wonky /CS trace for U103... my OCD is kicking in, that has to be straightened out in any later revisions

[agranlund]

On the Peripheral side of the 68150 you would be looking for issues with PD22 or PD30 but they are only going to through-hole components so it feels less likely.

[Oldskool]

Pin 38 of u106 is ground.

[danboid]

I know its already been suggested here at least once but wouldn't it be a good idea to start a new wiki page or two for the Raven?

Exxos, github or atari-wiki would seem to be the obvious first choices. I suppose its up to which one Anders prefers?

PS I'm not volunteering to document the Raven altough I would if I was building one. Maybe rev 2?

[agranlund]

Pin 38 of u106 is ground.

Bingo! Mystery solved, good stuff

[danboid]

Pin 38 of u106 is ground.

Bingo! Mystery solved, good stuff

No desktop pic no party!

[agranlund]

Bingo! Mystery solved, good stuff

No desktop pic no party!

Well you could build a cmdline only EmuTOS and play with it over the serial port.. sucks that you grounded a databit so you wont be able to run anything from an IDE drive though

Jokes aside, it feels like reflowing and inspecting those buffers should get you to the desktop
Those two are bastards to solder.