Homebrew clone of the OSI-440

[CommodoreZ]

So as folks have seen on my OSI-400 replica project, I've been spending plenty of time playing with the 440 video board, exploring what it can do.

As I've resumed work on my homebrew 6502 computer, the Cactus, I decided to take a crack at making a video card for it. To that end, I picked the 440 as a perfect starting point. Here's the result:

20240702_024125.jpg (3.38 MiB) Viewed 601 times

I changed a few things in the process:
Replaced 2102 SRAM with 2114's
Replaced Signetics 2513 character generator ROM with 2716/2816, quadrupling the available characters, populated with the standard OSI character ROM seen on later designs
Eliminated all provisions for graphics RAM and keyboard input
Simplified device select logic with modern parts, and DIP switches to change address to any 1K chunk within the standard address space.
Replaced video sync mixer with that of a 4066-based design intended for use on a PET for creating composite video.

20240702_024117.jpg (4 MiB) Viewed 601 times

And so far it seems to work with my old wooden prototype as well as my modern Cactus SBC design.

20240630_151715.jpg (2.68 MiB) Viewed 601 times

A few quick demonstrations of what the video looks like out of it. Sync isn't great (partly due to stray capacitance from the bodge-wires... I think)

graphics triangle sin.png (156.41 KiB) Viewed 601 times

graphics alphabet.png (134.34 KiB) Viewed 601 times

Anyway, I thought y'all might get a kick out this OSI-derived design.

[Thomas]

Nice board!
It looks like you have also replicated the rather odd video clock circuit made out of two LS123.
I would like to play with the potentiometers to see the effect on the monitor. There seems to be quite a good adjustment range.

Have you implemented the two optional jumpers in the 440 design, to stop the clock during HS and VS ?
To my understanding, these have the function to wait for the end of the HS/VS before the next characters are presented to the shift register.

If the HS blanking time gets increased, while the clock is adjusted up, there should be more characters visible on the screen by keeping the horizontal and vertical timing within NTSC spec. I do not have a 440 card, but I had a look recently into the schematic.


Thomas

[CommodoreZ]

Yep, I did not change the timing generators of the circuit.

I have plans to improve it with guard bands, and maybe improve the timing, while maintaining the 32x32 resolution, but with all of it visible on screen. Once that part is working, THEN I will add in all the bells, whistles, and feature upgrades. This is why I brought up the Progressive Computing mod here.

As for the clock stop on H-sync and V-sync, I did indeed implement that. I also added in jumper blocks to make it easy to pick which delay taps you used for clocking the shift register, and gating the output of that shift register. I could not find consistent implementations on 440's of those two signals, so I figured I would make it selectable. I've found that realistically you can't squeeze more than the 24x24 characters onto the display at any given time, no matter how much you play with the timing adjustments. My first attempt at building it resulted in multiple bodge wires to invert signals that I had gotten backwards. Inverted signals were the cause of every single initial problem on the board.

Anyway, here's the rest of the Cactus being demonstrated. I'm still perfecting a few things, not just the video device, but the front panel has been modernized and compacted from the original hand-wired design. The Rev B will hopefully fix the remaining bugs.

20240716_052136.jpg (3.59 MiB) Viewed 557 times