Ah, a fellow mechanical keyboard enjoyer! Welcome!
Are those XDA caps? Those are my favorites, too. (Although I've been using Kailh Choc switches a lot recently.)

The build looks really nice! I love all-in-one systems.

_________________
/Andrew

deck65 - 6502 slab with screen and keyboard | ПК-88 - SBC based on KM1810VM88 (Ukrainian i8088 clone) | leo80 - simple Z80 SBC
nice65 - 6502 assembly linter | My parts, footprints & 3D models for KiCad/FreeCAD

Nice looking machine and keyboard.

First of all, thanks everyone for the replies!

As for updates, I've designed and ordered a couple single IC test cartridges, since this is intended to be a cart based system after all. It will probably take a couple weeks for both them and the accompanying edge connectors to arrive.



I've also been considering some possible changes to the audio system. In particular, borrowing from a previous design of mine, which would be using a "wavetable" setup (big emphasis on those quotes), where each channel is assigned a wave defined by a single byte (so 8x1 resolution) that's held in a self-feeding shift-register clocked by the frequency generators. This would allow for a much greater variety in timbre compared to the current square wave PSG, 4 different types of pulse waves, and countless other funny shaped sounds. A much more unique and versatile sound system (I'm not sure if there's any historical soundchips that used this particular setup?), that as such, would be more appealing and interesting to write music for.

Though the above, of course, would come at a bigger IC count: 4 extra 74' chips.

Another possible change would be to replace the 82c54 that's used for note frequency generation with a Z80 CTC (Z84c30), this chip has worse frequency control, and would have to be run slightly out-of-spec, however, it also typically costs less than half of the 82c54 (if new) and is in stock more often. Considering the 82c54 is the most expensive chip in the design by far, I think this could be a worthwhile change to look into.

Finally, I'm going to be experimenting with using a transistor-based white noise generator to power the 4th channel. This is something I'll be testing using the current PCB hooked to a breaboard circuit. I'm very uncertain if that would even work (these types of circuits produce analog outputs, which would then be used to activate a CMOS binary input, very ugly), but it never hurts to test? The alternative would be to use a digital Linear-feedback shift register to generate the noise, but I'm not aware of a particularly simple and space efficient way to implement one using 74' chips. Or I could simply go the lazy route and just leave channel 4 as a bit-bang beeper, eh.



hmm not sure about the exact profile of these, and if they are XDA or not, but either way I rather enjoy lower profile keys, in fact the ones I've used for this build are leftovers from my main daily driver keyboard.



At the moment, no. As I personally don't feel it would add much to this particularly architecture/design, and would come at the cost of more complex address decoding logic.



Haha, well it should theoretically be able to do so. But first I'll have to finally get around to finally writing a memory monitor or fancier for the system
:p

Any reason not to actually use EEPROM-based wave tables? It's something I was considering and might work well with the multiplexing.


Could you just pass the onboard ROM's chip select through the card and back, so that if there's a card present it gets pulled high?

Interesting option. Sounds quite similar to the Namco WSG, which they used in their arcade games (here's a very good video that details exactly how that hardware works: https://www.youtube.com/watch?v=csFwHrQEdbg)

I thought of a potential way to implement something like that using the same number of ICs as my last proposal (I apologize for the crummy hand-drawn diagram):


In this case I'd remove stereo from the design and instead use one of the '670 chips to select the wavetable to use. Both `670 chips would hook-up directly to the ROM, with the 16 volume values being all precalculated in the ROM. Meanwhile `4050 chips (dual 4bit counters) are clocked by their respective oscillators to get a 4bit index into the wavetable.

To improve sound quality one could also place a D-type latch to buffer the ROM output, but I don't feel that would be necessary.

Edit: with this setup it might be possible to clock the inputs of the ch4 counter by combining the values from the 3 oscillators, to get a somewhat "random" percussive output.