Outstanding that you're following along with this.
There are always a certain number of people that can not, or don't want to work with SMD. The fine pitch components are especially tricky to work on. Also, through hole construction does lend some vintage authenticity to a vintage recreation. So although I did order a pile of SMD parts and have stock of both packages available, I am now leery of using them. The EVB board uses absolutely no SMD, and I think it would maintain that approach for as long as possible. I was only able to find the AS6C8008 by referencing the memory catalog, and it's not even very easy to find. By contrast, the 4008 is a wide 32 pin DIP, commonly available, and cheap as you say. It is always possible to make your own rules about any expansion boards, where for example you could pile on a few SMD memories to make a much bigger RAM disk if desired. If I'm not mistaken, quite a few OS disk images from that era, will still fit in 512k, even the 8 inch floppies were initially quite limited in capacity.
My Coco3 had 512k expansion, and the RAM drive software was quite impressive in operation!
I'm definite about using the 63B03RP, although the 6303Y might generate some "spin offs" later on. I will look up your Github account.
Another thing I overlooked, the HD6303RP appears to be pin compatible with the original Motorola part, the MC6803. Whereas, the HD6303Y has no Motorola counterpart AFAIK. I need to verify that in more detail.
Something I ran up against, outputs from NMOS devices are not CMOS level compatible, only TTL. So 74HCT logic and/or buffers may be needed in some cases, if the two logic types are mixed.
It turned out, there are some issues with the 6303RP that make it less appealing for this design. So I'm doubling down on the original design with 6303Y.
I made some changes that can be applied to a revised board, once I complete the game peripheral board. I piggybacked an address latch IC on top of a 512k by 8 NVRAM, so it's available in 16k pages where the original 32k by 8 NVRAM is located. That provides enough NV to make a small RAM disk in software. I built a baud rate generator board, which I had the foresight to support as a plug in module, visible next to the processor. With that, a 12MHz crystal can be used to run the CPU at 3.0MHz, the maximum. E clock stretching for peripheral access is already solved, Hitachi made that easy with an automatic option for memory ready signal MR. With that enabled, memory mapped peripherals are clocked at 1.0MHz.
Seen in the photo, the baud rate module to the upper left, piggyback NVRAM in middle, and a working proto board with the Philips SAA1099 stereo audio synthesizer on the right. The game board will consist of a TMS9918A video processor, SAA1099 audio, and an additional UART via an MC6850 or HD6350 ACIA. It's possible for the SAA and ACIA to share a common clock crystal with some simple baud rate divider.
The motivation for including a UART, is to allow the unit to run completely stand alone, with text/graphic display and also keyboard/mouse with one of these:
It's MC6801 object code compatible. I haven't gotten around to finding out whether it's a superset of the 6800 codes. I realized too late, disk images do exist for Flex etc, but those support 6800. Before, I hastily assumed that the code could somehow be ported. I have to look into that... today is rapidly disappearing though...
AFAIK, no MCU with on chip peripherals was ever made for 6809. However there are interesting variations in addition to the HD6309 (pin compatible), at least a PLCC version which I ordered and have on hand for future possibilities. That uses about half the board real estate.
