Notably, the RCA 1802 was also the processor found on the COSMAC ELF. Here’s the Wikipedia entry onthe 1802:
And here’s a thread on the HUG 1802, I found:
There’s a recent re-creation of the HUG 1802 that’s been showed off at various VCF events: cskordis ETI-660-Learners-Computer. (ETI-660 was the project’s name in Australia.)
Good luck getting the CDP1864 colour graphics chip, but there are variations available that use (slightly) more common silicon.
It’s yet another project from those busy folks at the COSMAC ELF Group.
Hmm, so taking a gander at the schematic has me wondering. I’ve always liked the idea of using a decoded N-output to select a big peripheral chip. However, I noticed the R/W* line on the 6821 here is hooked to the A2 address line. So if an OUT 2 is done and you have a sw error due to incorrectly Set-Xing R(x) or haven’t kept up with the correct housekeeping within the correct R(x), then a high A2 will allow the 6821 input to collide the data bus with the OUT memory read during the TB high pulse. Maybe I’m not understanding this timing correctly. Any 1802 fans here agree with this? Of course, if so, I’d never make a sw mistake like that on my first love uP!
In the dim dark distant past a friend and I published an article on hooking up a baudot teletype printer to the ETI-660 (surplus, so cheaper than a “real” printer). Although it looks like a lot of the ETI articles are online, this ended up being printed in a special one off computing magazine that ETI produced, so I can’t see any references. Shame, only time I was ever a published author. 
I wonder if this could be PAL-ified?
I think this has been discussed before here, but the 1802 has excellent die shots on visual6502.org: RCA 1802
These were used for a complete reverse engineering job in the now-inaccessible visual6502 wiki which I have resuscitated here: v6502demo | Demo of recovering and reformatting some CC3.0 licensed content
But isn’t the 1861 black and white, where the PAL 1864 is colour?
Fair point; I misremembered those two parts as being mostly equivalent (and now looking at the datasheets, the 1861 is 24 pin and the 1864 is 40 pin). Ah well, more decaying neurons uncovered.
I do remember disassembling the ROM in my youth, and being surprised that it was partially written in CHIP-8 itself - almost a meta-circular interpreter.