LFSR is 15bits on TI chip and has taps on bits 0 and 1, white noise has lower periodicity and is mellower in general, 15 bits also means that pulse mode has 6.6666...% duty cycle and also shifted pitch. VDP LFSR is 16bit with taps on bits 0 and 3, resulting in more longer and more "percussive" white noise and pulse mode duty cycle is 6.25% with no pitch shift compared to other channels.
Tone channel counters count down on TI chip and up on VDP, implication is that freq value of 0 produces lowest possible freq (1024) on TI while on VDP, values 0 and 1 both produce same result (1). Games often use freq value of 0 as means to silence the output in Sega stuff, which on TI will produce 110Hz tone rather than inaudible ultrasonic freq (which is eaten by LPFs).

Thanks for such thorough explanation. Sega’s variant should go in the card for sure, but in which form. 2x PLCC84s do not seem so scary with time) Are they 5v tolerant? And what do you think about that opamp influence on sound Jo22 was talking about?

These CPLDs are 5V so no problem. A more modern chip such as MaxII (3.3V only and only in QFP, no PLCC exists) would be able to fit everything into single chip. Some bigger Max7000S chip could also but they are only in QFP too and I have got the impression you would rather not use such parts and have through hole or at least sockets. Anyway the design isn't completely finished or tested and right now my hands are full with some other things...

Sound of MD1 is largely shaped by the super low quality chip used as headphone amp, and most particularly its misuse. It is wired in a very strange way there. in any case you can simulate the sound with heavy LPF (matter of changing capacitor value in the feedback loop) and perhaps use of low quality opamps at high gain so that you get all the distortion that exists in original circuit.

Thank you for the explanation, Tiido! ^^ I once read about this weird amp configuration in some Sega forum, but lost track about it.
The reason I mentioned it, is, because the default connection to the TV set was RF (antenna cable),
which gave a different sound than the headphones output. So that's why it came to my mind.

Some games apparently were meant to sound right via RF modulator (-which I thought was fed by the mono amp, too-) in first place,
so they may not sound nice/as intended if played back "unfiltered" (-akin to the dithering "issue", where some games don't look right if not played via RF or Composite-),
while other games perhaps sounded better/as intended in stereo via headphones jack.

Anyway, I never really had a chance yet to install that clear sound mod into one of my MD1 models. 🙁

Edit: I'm sorry for the bad writing. I have difficulties to put my thoughts into words.
In essence, I was trying to say, that the filtered sound should be taken into account when making the card.
I don't mean that the sound should be low-quality by design, but maybe there should be some testing with existing MD soundtracks
before finalizing the card. Maybe then a simple low-pass filter can be added, if necessary. It could be enable/disable by a simple port write perhaps.
Maybe implemented as a simple flip-flop. Some early devices, like the Amiga or SB Pro also had fixed filters, I recall.

Edit: That being said, these are just ideas that come to mind. They are in no way binding, of course.
I think this a facinating project, no matter how the final card will come out in detail. 😀

Does the Chip Video too ?

EDIT:
Indeed! can we use this on PC Card also ?

Ok, here is the plan: to make a prototype asap (say, order 5 pcbs on JLCPCB) with YM2612 and a TI PSG for software development on a working card, h/w tests etc. I'll take one of them and send out other 4 to anyone who is interested and wants to contribute in hardware or software parts. After that I hope it will be clear how to integrate VDP's PSG into the card, in which form, and then I will make the final card and release its files and any software written to that moment (at least a VGM player). As I understand Tiido and Jo22 are in, who else?

Hm i`m also interested but its unlikely i will be able to contribute something useful.
As i`m less experienced and fight my own building site 😀

Maybe for the PSG Site i send a prototyp to lotosdrache and he tries to develope a driver for games to make them working on AT Class Machines. And it looks like some times games also need patches for timer routines..
I`m not sure if it will help here 😀

He already did a New Driver for OPTI ISA Sound Cards.

^ what's this about a new driver now???

Well, it seems like I'm not the only one who dreams about Sega MD sound on a PC. 🤣
I've made a mockup of PSG+OPN card some 20 days ago. But it is in the frame of my FMonster project, so PSG will be indeed of TI variant, no place for everything else. And DAC functions have to be emulated through 8-bit PCM output via Covox Speech Thing-compatible DAC.

LABS did you try to assign port addresses to parts of your future board?
As for the FMonster, PSG can be at 0x1C0/0x1E0/0x2C0 (compatible with Lo-tech Tandy Compatible) and takes 2 ports from base.

OPN2 I've selected to be on one of these: 0x2A0/0x2B0/0x2D0/0x2E0 and it takes 4 ports from base.

I'm not insisting on your board using the same addresses, but if it could, it would be great: new software can be compatible both with FMosnter and your GENIE. 🙄
I can share schematics for address decoders on 74HC138 for both PSG and OPN2 (each takes 2x 74138).

Either way I'll keep an eye on your project and probably buy a board when it will be ready. So that I can get "true" Sega MD sound, not like half-emulated on the FMonster with TI PSG and 8-bit DAC. Keep it up! 😎

Heh, there is one saying.. 🤣

I did not decide on the ports yet, but the idea was to use 500h-50Eh, with bits 2 and 3 selectable with a dip switch to select 500h/504h/508h or 50Ch, using 688 comparator and a 138 to decode 10 MSB of the address. A0 and A1 will select OPN2 (01b & 10b) or a PSG (11b).

If you are putting a TI PSG variant, it make no sense to do the same for me. So Tiido's idea to use VDP PSG should be implemented in GENIE somehow.

When are you planning to make PCBs? I'll need one for software development.

LABS wrote:

When are you planning to make PCBs? I'll need one for software development.

Debug block of PSG is ready and I already have the boards. But I didn't test them yet.
OPN2 block is not done in PCB yet but is ready to be ordered.
I can order OPN2 boards and then send you PSG + OPN2 debug boards.

LABS wrote:

I did not decide on the ports yet, but the idea was to use 500h-50Eh

This is gonna produce a problem with cards doing 10 bit address decode, 500...50F range will conflict with gameport due to mirroring from 10 bit decode of some other card. Something to be aware of.

Tiido wrote:

LABS wrote:

I did not decide on the ports yet, but the idea was to use 500h-50Eh

This is gonna produce a problem with cards doing 10 bit address decode, 500...50F range will conflict with gameport due to mirroring from 10 bit decode of some other card. Something to be aware of.

Ah, ok.
Which region is safe to use?

leftovers from 2xx and 3xx range are the usual places. 3xx range has more competition since some base hardware lives there and many network, MIDI and other cards put something in there. 2xx is less crowded.

Here's addresses of standard things found in there, many which are off limits (add 400, 800, C00 to see mirrors, which are also offlimits) :

200...20F - Gameport
278...27F - LPT2
279 - ISA PnP (off limits or PnP cards will not work), purposefully shared with LPT2 range.
2E8...2EF - COM4
2F8...2FF - COM2
376 - IDE controller (off limits)
378..37F - LPT1
3B0...3BB - Video (off limits)
3BC...3BF - LPT3
3C0...3DF - Video (off limits)
3E8...3EF - COM3
3F0...3F7 - Floppy+IDE (off limits)
3F8...3FF - COM1

Rest of the stuff includes various sound cards, network cards and other misc cards.

Speaking of I/O addresses, over in PC-98 land, the OPN/OPNA chip on the sound boards used for those machines generally use I/O port 188h, but again, that's on very different hardware from IBM and compatibles.