...continued.

In r0, the 74hc273 (U4 for those following along on the schematic), was hooked up as a very basic debugging output, outputting everything written to the IO range of $4000-7FFF. 

This was good enough for figuring out if the 6502 was running or not, but not really very useful. 

In r1 the complexity has increased slightly as I used the leftover gates I had to make it into a write-only register in the range of $5000-5FFF, that's not used for anything else at the moment. This means that any read or write in that range will set the register to whatever is on the data bus. Though not optimal, certainly better than before. 

This means it will work as a cheap output register that can be used for simple control signals. With this upgrade, I decided to use the first bit as a video enable line, since it will protect the connected screen from spurious output during boot. It goes through an inverter and to the chip enable of the 74'165. I also put a "small" capacitor on the ~CE - too large I think, because a cold boot takes a few seconds instead of the few milliseconds I was aiming for. Probably 10nF is about the right range to delay video output until clocks are stable and VRAM is cleared, though I still wouldn't connect an old CRT to it. Or maybe I would.. But maybe hold off trying yours :)

The next two control lines I decided to use as banking lines for the character ROM. I can then define a second, third and fourth character set, 128 characters wide. So far I haven't put this to use but I can imagine using it for graphics, like PETSCII, all graphics characters or another font. This still only uses 4k of the 64k available on the W27c512 I use as a character ROM, but considering I got the idea from Don Lancasters 1978 "Cheap video cookbook" when ROMs were still available in 32 byte packages and dedicated character ROMs were still a thing, all I can say is ... can I still get cheap 1, 4, 8k PROMs somewhere?  The w27c512's from Aliexpress are less than a dollar each. As far as I can tell they are pulled from random scrap - and not even wiped before sale. The downside for these ROMs are that they need a dedicated programmer like the TL866II+ (Arduino ROM programmer won't work, since these need 14 volts for programming) and they take a bit longer to write. About 30 seconds for the whole thing. 

Speaking of ROM, I use the SST39SF010 for development since it's only about 5 seconds to write - this is also the explanation for the 32 pin ZIF socket and J5 Jumper, which lets me easily change between the W27c512 and the SST39SF010. 
This is another upgrade from r0 since that actually only had a DIP package landing pattern, meaning I couldn't fit the ZIF socket even though the space was available - the holes were simply too small (doh!). Fixed in r1!

But what about the reset and buffering? Find out in the next log :) 

Here: https://hackaday.io/project/184725/log/204911