#composite video

The Cactus Video Card Mk II Rev D seems to have fixed most of the bugs present in previous revisions, but it is far from perfect.

With HCT series parts installed, the video locks up cleanly to all of my monitors, both color and monochrome alike. It also looks good in both alphanumeric and graphics modes alike.

However, once I install LS series parts...

The duration of the inverted sync lines during the Vertical Blanking Interrupt is unstable. It jitters from 7 to sometimes as many as 19 pulses (possibly more), which causes flickering video.

Doing some digging I found that with HCT parts, VBI was only 1 line long, not 6 like I had planned. I was able to make one bodge and change that back to 7, which isn't great but it is better. Then we 3 more bodges, I got it to an even 6, without cutting off the last line of video early.

I will have to make these bodges more permanent and then I can test again with LS series parts to see if I cleaned up any of the problems. Worst case scenario, I have to use this with HCT parts only.

So the Cactus Video Card Mk II Rev C is beyond help.

I built it out in the minimal test configuration to verify functionality, and the switch from 74193 counters to 74163 synchronous counters fucked everything up, causing a myriad of issues that I did not anticipate. While some of these could have been made obvious through simulation, I missed the bulk of them.

Counting up requires additional precise qualification of the clock inputs, which I failed to account for exactly -- synchronous counters are a pain.

I also was used to my carry-out pin being a terminal count indicating that it had finished counting to 15, which is not how the '163 works.

Lastly, the reset signals are synchronous too, and while the 74161 alleviates some of these issues, it wasn't going to be a panacea.

The pinouts between the two are so radically different that I decided it was going to be better to revert to the old counters.

Three different revisions of the Cactus Video Card Mk II

Debugging of the Cactus Video Card Mk II Rev B continues.

The last issue remaining with this design is fixing the inconsistent reset of the video state machine. About 10-20% of the time now, either on power on (though more likely from pressing the reset switch), the state machine will end up in a bad state and output incorrect video, requiring another reset before it settles in the expected pattern.

New toys for the video plant.

On the left, a comb filter (lets you convert between composite and s-video)

Bodges have been added to fix some bugs, which now means that the primary functions all seem to be working.

Check out that 6810 SRAM chip from 1975 now given purpose, probably for the first time since it was manufactured. All 128 bytes of RAM used for custom character graphics. You can see those on the second row.