32768 color dual Shifter ST mod - schematics lost?

[sporniket]

A little morning edit :

• Renumbered the bits of the output, to have main shifter output MSBs of R/V/B channel, and the auxiliary shifter output the LSBs .
• Added a way to test whether the address range of palette registers is 0—15 or 16—31.

rgb565-experiment.pdf

(100.28 KiB) Downloaded 127 times

[sporniket]

I did the color Digital-Analogic Converter, and I understood how it re-injected the result into the motherboard.

The schematic is almost complete (I have to deal with power sources so that Kicad ERC will be happy), and unless there is an error, that's it.

rgb565-experiment.pdf

(142.5 KiB) Downloaded 110 times

I have put the project on github : https://github.com/sporniket/atari-st-m ... experiment

(Please note that the libs in kicad-symbols-generated must be generated before opening the project to avoid some error messages)

[mrbombermillzy]

Great work!

I really appreciate you tackling this!

[sporniket]

Thanks. There is still something wrong, because I felt the resistor values too strong in the initial mod. On second thought, I will revert to those values, as the one I chose may give a resulting current that may satured the transistors that amplify each color component signal (meaning my choice leading to a washed, too bright pictures at best, and hurting the monitor at worst).

For the record, I chose my initial resistor values to have a slightly higher equivalent resistance than the stock resistor values in the motherboard (basically 1K, 2K, 4K in parallel).

Anyway I edited the values. The github is up to date, and here is the fixed schematic.

[sporniket]

Progress... the github repository is up to date.

Noteworthy : 

• I provisionned a pull-up to match the one on the data bus.
• I use through-hole only, to lower the difficulty of assembly, and at times it's a bit tricky.
• It will be a 4-layers PCB, with a ground plane and a power plane.

2021-11-17--preview-wip.jpg (128.52 KiB) Viewed 3180 times

[DoG]

Progress... the github repository is up to date.

Is it going to fit in the shifter 'can' in the middle of a STfm board or is the MegaST you are aiming for or just proof of concept?

[sporniket]

Regarding the pcb size :

@DoG, I believe that it will fit. For now I estimated what I could do using my own pictures of this enclosure (image 12 of 17).
And I will take time to take out the STF and open it and check with the pcb printed on paper.

About my aim :

I had looked into it out of curiosity, and then I end up saying that I would try to make a proper schematic (done), and maybe a pcb (doing). To me it will stay a virtual exercise (I will manufacture a batch for checking the real thing though), and @mrbombermillzy is interested to test it. The main problem that I expect (apart blowing an healthy computer, so I try to avoid shorts) is timing problems, I also have doubt whether the cristal oscillator is able to drive 2 Shifter at the same time, and then usual problem with noise and ringing signals.

And I had to come with my own way, as I didn't fully grasped how the original author used the quad-channel tranceiver and the octal one. I use 2 octal-channel transceiver that are mutually exclusive. And I prevent the second shifter to drive the bus.

[mrbombermillzy]

Great effort @sporniket

What we really need BEFORE you get a batch of boards ordered is really a 2nd (or more is better) set of eyes to check over the circuit.

I would happily help, but my digital electronics circuit design skills are pretty lousy.

Perhaps @Smonson or @troed (or any other 'shifter HW theory gurus') can kindly help?

I would HATE to see you waste the effort if the boards arent right and cant be bodged easily.

[Icky]

Great effort @sporniket

What we really need BEFORE you get a batch of boards ordered is really a 2nd (or more is better) set of eyes to check over the circuit.

I would happily help, but my digital electronics circuit design skills are pretty lousy.

Perhaps @Smonson or @troed (or any other 'shifter HW theory gurus') can kindly help?

I would HATE to see you waste the effort if the boards arent right and cant be bodged easily.

One option is to make the Phoenix Shifter core do this. I'll see if I can modify it to replicate your circuit @sporniket. Unfortunately I am not able to test as all my Atari stuff is in storage still.

[sporniket]

Hello all,

I will gladly accept any feedback. In the meantime, here is the completed "experimental release candidate #1" (experimental, because of the ability to select wich range of adresses will have bit swapped) :

rgb565-experiment--v1.0.0-xprc1.pdf

(159.31 KiB) Downloaded 87 times

The github is up to date.

2021-12-18--almost-ready-recto.jpg (155.42 KiB) Viewed 3110 times

2021-12-18--almost-ready-verso.jpg (116.66 KiB) Viewed 3110 times