I think the simplest solution to try it out is to use a dedicated atmega328 just for the FIFO. It receives OPL3 command at bus speed, puts it into 2k RAM queue, and then sends it out to OPL3 when time comes by controlling OPL3 pins with appropriate timings. 2k will be more than enough for that, probably we don’t even need to check it for overflow and touch /IOCHRDY to slow things down, what makes things even simpler. I already kinda did this in my OPi project-USB 2.0 to OPL3 sound card, where OPL3 commands were sent over USB in packets to atmega328 and then passed to OPL3 with correct timings. It worked flawlessly from DosBox.

I would have thought that the MCU will have some difficulty receiving the commands at ISA speed, but then again, ISA is very slow and there will not be large sustained transfers anyway. That's definitely viable then ~

The project feels very refreshed 😀

Finally finished the prototype

It took lots of time, but now it performs much better and smoother than the original SB2.0 overall, supports wider range of CPU speeds and still use common components.

Audio performance is even better than Blasterboard's: -100 dB RMS !! Dead silent

Packed all that into a new (and hopefully final) prototype

Thinking about base IO port selection. SB8 now has 200h / 220h / 240h / 260h. Are 200h and 260h good "extras"? Never used anything other that 220h actually.

Excellent results, awesome ~

$200...20F conflicts with gameport which is why basically nothing gives it as an option. $260 should be ok.

LABS wrote on 2024-08-22, 18:53:

Thinking about base IO port selection. SB8 now has 200h / 220h / 240h / 260h. Are 200h and 260h good "extras"? Never used anything other that 220h actually.

Before SBPro (which added a CD interface at 230/250), only 16 (10h) ports were used. 210/220/230/240 or 220/230/240/250 sound like a more sensible option for an SB 2.0 clone than 200/220/240/260, especially due to the game port conflict at 200. According to the original Creative Labs hardware documentation, SB 1.5 offered 210..260 in increments of 10 (i.e. 6 choices), while SB 2.0 already narrowed down the choices to 220/240. As your target is SB 2.0 compatiblity, everything except 220/240 is a bonus compared to the original hardware.

Tiido
mkarcher
Thanks for pointing that out! Completely forgot about the joystick port 🤦

Really cool project.

But if you've upgraded the original SB 2.0 design with an OPL3 synthesizer anyway, why not go a step further and add Sound Blaster Pro 2 compatibility? Without any of the CD-ROM interface stuff, I mean. Just to have stereo support (and actual OPL3 support) in the games that supported it. Not to belittle your impressive work so far. It just seems like a trivial extra step in comparison to the work already done.

Thanks for considering!

digger wrote on 2024-08-24, 13:03:

... Just to have stereo support ... seems like a trivial extra step ..

It seems trivial at first, but adding support for stereo requires implementing the entire SB Pro mixer chip, which is not a trivial task for the discreet circuitry. But adding support only for a single mixer register with "stereo enable" flag is not enough, because games which use Miles Sound System library for example (there are hundreds of them), use mixer calls for SBPro verification even in manual settings mode. Implementing mixer will require some kind of CPLD or FPGA chip and then all the digital schematic should be reimplemented inside this chip as well. But then it will be a totally different project.

Watching this space with great interest!

I agree, do not blow up the scope of this project. Will there be an option to assemble your own card? I know it will be few people who want to do that, but I like doing SMD soldering by now, and this would be a great piece to practice on. I still have the original BlasterBoard here, even if it's currently not in use!
It has been a very long time since I last soldered a sound card, so I would be definitely down for a kit. 😀

root42 wrote on 2024-08-27, 06:51:

...Will there be an option to assemble your own card?
...I like doing SMD soldering by now..

Not planning the kits this time - too much struggle with handling the components, but I could make one for you if you like

LABS wrote on 2024-08-27, 21:21:

root42 wrote on 2024-08-27, 06:51:

...Will there be an option to assemble your own card?
...I like doing SMD soldering by now..

Not planning the kits this time - too much struggle with handling the components, but I could make one for you if you like

I wouldn't say no...

LABS wrote on 2024-08-25, 16:17:

digger wrote on 2024-08-24, 13:03:

... Just to have stereo support ... seems like a trivial extra step ..

It seems trivial at first, but adding support for stereo requires implementing the entire SB Pro mixer chip, which is not a trivial task for the discreet circuitry. But adding support only for a single mixer register with "stereo enable" flag is not enough, because games which use Miles Sound System library for example (there are hundreds of them), use mixer calls for SBPro verification even in manual settings mode. Implementing mixer will require some kind of CPLD or FPGA chip and then all the digital schematic should be reimplemented inside this chip as well. But then it will be a totally different project.

Completely fair. Was there any OPL3-based sound card back in the day with stereo support that would be trivial to implement on top of basic Sound Blaster 2.0 compatibility?

digger wrote on 2024-09-02, 12:32:

Was there any OPL3-based sound card back in the day with stereo support that would be trivial to implement on top of basic Sound Blaster 2.0 compatibility?

None as far as I know 😀

The main idea behind using OPL3 is to run games which refuse to sound normally on machines faster than a 286/386. For example - Prince of Persia sounds fine now on my Pentuim 166MMX, but it played FM garbage on the same machine with the original Sound Blaster 2.0. And it is now also possible to hear a complete OPL3 score in games which support it - I just select Sound Blaster 2.0 for sfx and Sound Blaster Pro II for music in setup - Warcrafts, Gabriel Knight, Descents, etc etc... And also it was quite trivial to change OPL2 to OPL3, while keeping the schematic discreet and simple

Fresh cards for testing 😀

LABS wrote on 2024-09-02, 15:50:

Fresh cards for testing 😀

mmmmmm
tasty~!
😁

Nom nom. 😜

Looks good in "AsusTek" yellow. 😀

I'm wondering about games compatibility. Since there weren't any SB2.0 with OPL3 back in the days auto detection might get confused. So game must have dedicated OPL3 option for music playback or special variable set. This recent thread is about OPL3 games - could be good test set.

Edit: Never mind - PAS16 has both Thunderboard(SB2.0) and OPL3 chips.

LABS wrote on 2024-09-02, 15:50:

Fresh cards for testing :) […]
Show full quote

Fresh cards for testing 😀

The attachment sb8x4.jpg is no longer available

The attachment sb8-p1-1.jpg is no longer available

The attachment sb8-p1-2.jpg is no longer available

The attachment sb8-p1-3.jpg is no longer available

Oooooooohhhhhhhhhh...... Right there...

vutt wrote on 2024-09-02, 18:07:

Edit: Never mind - PAS16 has both Thunderboard(SB2.0) and OPL3 chips.

Hi. Fascinating topic! We could think of it more like an SB 1.5 with extras, maybe. It reports DSP 2.00, I think.
Thunderboard was like an less noisy, slightly enhanced Sound Blaster with extra hardware-based compression types and higher sampling rate.
Both SB 1.5 and SB 2 had DSP versions with auto-init DMA support (DSP v2.00, v2.01), so the most important SB stuff was covered on Thunderboard, eitherway. 😀
Things like high-speed mode required SB 2.0/DSP 2.01; not sure if Thunderboard can do that.