32768 color dual Shifter ST mod - schematics lost?

[troed]

I've just read through the well known textfile describing "how" to add a 2nd Shifter to a regular ST and gain not only STE color palette compatibility but increasing it all the way to 32768 colors:

http://atari4ever.free.fr/hardware/zip/32kcolor.zip

Cool, I think, so I wanted to look into possibly doing it to one of my machines. However, there's no schematic with the textfile, only a parts list, and I've since seen others trying to hunt down the schematics (or the original designer, Barry Orlando) over the years without success.

Posting here in the hopes that somewhere someone has a copy of those schematics that could be scanned.

/Troed

[Smonson]

I think we could recreate equivalent schematics fairly easily from the BOM. Basically we are talking about a board that plugs into the shifter socket and connects two shifter ICs together, sharing all the control signals. The data bus will need to connected straight through while normal pixel data is being latched, but while CS is asserted, they would each see different bus lines.

I think most likely one of the shifters will be always wired up to the stock 9 colour bits, while the other shifter sees the 6 new bits, transposed into the positions of the original bits.

Then 3 new 5-bit DACs are made out of all those resistors.

Are you gonna build one? :)

[mrbombermillzy]

Thanks for the heads up Troed! I must of missed this somehow. Well, I'm gonna bump it and see if anyone bites ;)

Can someone pretty please have a go at this?

From what work I've been doing with graphic manipulation, a bigger palette can make more difference than number of colours on screen in some cases.

Would be almost like Amiga AGA 16 colour pictures... and that's without the help of Photochrome!

[sporniket]

A small necro-bump, because the topic spurred between me and @mrbombermillzy on AF.

A few monthes ago, for the sake of curiosity and killing some time, I have read the description of the mod, and did a doodle :

2021-12-07--rgb656-mod-doodle.jpg

I also wrote down a google doc, that I copy here for the sake of completeness :

RGB-565 color mode
with supplemental ST Shifter
Abstract
Description of adding a second ST shifter to get finer color definition.
When writing into color palette register the following value 16 bit value :

Code: Select all

GRGBRRRRGGGGBBBB
5444032103210321

Then the first shifter receive :

Code: Select all

00000RRR0GGG0BBB
     321 321 321

Then the second shifter receive :

Code: Select all

00000R000GG00B00
     4   45  4  

Then a Falcon-type DAC receive :

Code: Select all

R[7..0]=1R2,1R1,1R0,2R2,2R1,2R0,0,0
G[7..0]=1G2,1G1,1G0,2G2,2G1,2G0,0,0
B[7..0]=1B2,1B1,1B0,2B2,2B1,2B0,0,0

Testing
In color mode

• Write a program displaying "333 444 555 565 "
• Display 4 rectangle, each one under one label, with color 1,2,3,4 respectively
• generate the RGB3, RGB4, RGB5 and RGB565 16bits color encoding for a grey level from 0 to 63 (6bits).
• For each couple of scanline starting from line 8, change the palette registries to set color 1, 2, 3 and 4 to the grey color in RGB3, RGB4, RGB5,
• RGB565 respectively, starting with grey level 0
• Continue the trick until key pressed

that said, I believe that this rgb-656 could be implemented inside the fpga recreation of the Shifter done by @Icky here : https://www.exxosforum.co.uk/forum/viewt ... 630#p71945

[Icky]

A small necro-bump, because the topic spurred between me and @mrbombermillzy on AF.

A few monthes ago, for the sake of curiosity and killing some time, I have read the description of the mod, and did a doodle :

that said, I believe that this rgb-656 could be implemented inside the fpga recreation of the Shifter done by @Icky here : https://www.exxosforum.co.uk/forum/viewt ... 630#p71945

@sporniket once we have a completely working shifter recreation of @Smonson's brilliant efforts then improvements are possible.

[Elliot]

I just posted about this on another forum, I remember this upgrade. I never understood why/how a second Shifter was required! Max colours was still 16 right but from are larger pallet. Was the second Shifter just a name or was it literally another ST Shifter? If that is the case what did it do? I assume there was no extra BUS access, the pallet was still (the same) two bytes per colour so did the other Shifter take the fist Shifter output and add another shade or...

[Elliot]

I am sure there is zero point as nothing would support it. Maybe, maybe there is a paint package that does but...

The hardware may be compatible with some STE software but I suspect any software that could potentially use it would not as it would check the system type first.

[Smonson]

I have an incomplete understanding also, but my interpretation of how it would have worked is that during writes to the palette registers, the second shifter would receive the data bus lines transposed like this:

Code: Select all

D14 -> D8 (red bit 5 -> shifter 2's red bit 1)
D13 -> D4 (green bit 5 -> shifter 2's green bit 1)
D12 -> D0 (blue bit 5 -> shifter 2's blue bit 1)
D11 -> D9 (red bit 4 -> shifter 2's red bit 0)
D7 -> D5 (green bit 4 -> shifter 2's green bit 0)
D3 -> D1 (blue bit 4 -> shifter 2's blue bit 0)

I *think* this is the same as @sporniket 's diagram.

with all other bits set to 0. At all other times, it would see the data lines normally.
The parts list includes a 2-input NAND gate (I didn't do the maths but I assume that would be enough to decode the palette write condition), an octal bus transceiver (to drive the six transposed signals) and a quad bus transceiver (to drive the three zeros in the unused parts of the palette).

During a pixel output the two shifters combined will then produce a 6-bits-per-colour output instead of the normal 3-bits-per-colour, which would normally be driven through a 3-bits-per-colour resistor DAC to convert to analogue on the ST mainboard, but in this case they're all routed to a new 6-bits-per-colour DAC here, and each of the resulting 3 analog colour outputs are injected into the mainboard using flying wires.

The only thing I couldn't work out is how the normal data lines are driven to the second shifter when NOT writing to the palette, but there aren't quite enough transcievers in the parts list for that, so... ¯\_(ツ)_/¯

I agree with everyone else who already said this is pretty useless, with 16 colours there's not much call for a gigantic palette.

[Darklord]

Agree about the 16 colors and mostly useless gain, but...

My understanding is that anything that takes advantage of the
increased palette of the STe will also work with this mod.

For example, Photochrome.

This is the same concept as the John Russell Industries 4096 color
upgrade, right? JRI 4096C board.

https://info-coach.fr/atari/hardware/memory.php

Go down the page a bit and you'll find some pictures...

JRI-shifter-conn.jpg

JRI-connect.jpg

JRI-installed.jpg

[Smonson]

This is the same concept as the John Russell Industries 4096 color
upgrade, right? JRI 4096C board.

I assume so. I've never seen any info about how it works.