Tiido wrote:

Simple linear interpolation will fail badly with FM output, all the noise type sounds (such as precussion) will sound very different from my experience, elaborate methods are required to get output that doesn't add any flavor and it can be hard to achieve.

Any chance you can point me at proper direction ?

Tiido wrote:

SB and WSS sample playback isn't going to cause any significant headaches though.

OK, it's exactly what I was playing with (SB DSP) - will try OPL2 to compare...

Tiido wrote:

That chip seems to be decent enough, though I would still look at some PC targetted codec chip as they have better dynamic range (5V analog side rather than 3.something) and more inputs (so you can handle CD and other things too without more analog stuff).

Actually, not so easy to find anything similar - more inputs require a separate analog mixer (at least, I couldn't find anything better)...

Why implement a soft core when you can purchase something like a Zynq?

Sinc based resampling will produce the cleanest output with minimal distortions, depending on how many taps you'll be using, the more the merrier but at increasingly diminishing advantages. Some sort of tradeoff will eventually be reached as you aren't going to be able to do infinite amount of multiplications per sample. Doing some muls in parallel and some in pipeline fashion is going to relax resource and timing tradeoffs.

TI makes bunch of things with many more analog inputs and low cost too. Seems stuff with 5V analog side are rare/expensive. TLV320AIC3204 caught my eye.

SoC FPGAs could be useful, though they're lot more expensive and quite a bit overkill too. You also need DDR memories and whatnot... though you do gain the possibility to implement a full MIDI softsynth on one.

Tiido wrote:

TLV320AIC3204 caught my eye.

Looks very nice ! For now will stay with 8731 (have some experience with it), but use TLV if need more analog inputs.
Thanks !

There was (succesfull) attempt to recreate Sound Blaster 1.0 in the past. Unfortunately documentation is lost.
http://www.deep-shadows.com/hax/wordpress/?page_id=364

zyga64 wrote:

There was (succesfull) attempt to recreate Sound Blaster 1.0 in the past. Unfortunately documentation is lost.
http://www.deep-shadows.com/hax/wordpress/?page_id=364

Thanks for the link! I've seen it before, but it still amazes me. It looks beyond cool, it's awesome. 😁
Like a piece of technology crafted in a post-apocalyptic age. Also, great respect for the wiring.
Must have required lots of patience to assemble it the way it had been done.

There is another project from the same time, but a bit different approach - it’s very close to the original SB, uses Russian clone of 8051.
BTW, I have completed and tested a firmware for all DSP 2.0 commands (no ADC part yet, just DAC). Everything is written without any NIOS-specific things, so should be quite easy to port to different CPU/MPU. Probably, will re-write everything completely - already gained some C experience, so can see how terrible the code is 😀
Unfortunately, have to take a break for a couple of weeks - wind is blowing, kite is calling 😀

The 8051/8042 family is very interesting. When I started to learn about with MCUs/µCs, I read about that legendary & unobtainable 8052AH-BASIC chip.
This was before I learned about the PIC16C84 (later 16F84)), which I tinkered with for many years and also build a
custom serial programmer for PonyProg - Yeah, pony-prog.. Like.. "pony" :

Speaking of the 8042, I heard someone wrote a full-fledged PC emulator (was it 8086tiny ?) for a very fast descendant of the 8042µC.
To think of, that a PC runs inside of what once merely was a keyboard controller, still baffles me! 😕

Anyway, back to the sound cards topic.. Since the 8051 was also used in SB clones, I wonder if they did urun on -uhm- original code.
To me, the situation is quite a bit comparable to what happened in the 80s with the many XT clones.
Some of them had their own firmware, like that anonymous PC BIOS. That's why I wonder if some of the many
no-name SB 1.x/2.x clones had been examined in this regard (Quickshot, Protac, etc).

https://www.fleasystems.com/flea86.html

Flea86 is 80C51 running 80186 Emulation.

Hm when using a FPGA would it not a better aproach to make it universal ?
Supporting Cores like Mist.

Eg Soundblaster Core, GuS Core, WSS Core and so on...

matze79 wrote:

https://www.fleasystems.com/flea86.html

Flea86 is 80C51 running 80186 Emulation.

Ah, I see. Thanks for pointing that out. I only remembered it must have been something "little". 😅

Perhaps some parameters should be outlined. Is the objective to duplicate the hardware as closely as possible or is the priority on duplicating software compatibility and audio output (with select improvements) irrespective of how this is accomplished?

gdjacobs wrote:

Is the objective to duplicate the hardware as closely as possible or is the priority on duplicating software compatibility and audio output (with select improvements) irrespective of how this is accomplished?

Exactly... Personally for me just making an accurate replica of SB is not interesting. And it could be easily done in a few days (taking into account the time I have spent already).
So I'm playing with the idea of "universal" 8-bit sound card on modern components.

newold86 wrote:

gdjacobs wrote:

Is the objective to duplicate the hardware as closely as possible or is the priority on duplicating software compatibility and audio output (with select improvements) irrespective of how this is accomplished?

Exactly... Personally for me just making an accurate replica of SB is not interesting. And it could be easily done in a few days (taking into account the time I have spent already).
So I'm playing with the idea of "universal" 8-bit sound card on modern components.

I'd like to make a distinction between the "music" part and the DAC part. For the music part, personally I would like to see original hardware used as much as possible, particularly the OPL3 chips and the Tandy compatible sound chips. And we wouldn't have to start from scratch here. We could simply "piggy-back" on other existing open source retro sound card projects such as the Resound OPL3 card and the Tandy 1000 ISA sound card. For implementing the MPU-401 compatibility, perhaps we could (at least in part) adopt and integrate the design from the HardMPU project.

That would leave only the DAC part for us to newly implement. For this, I'm completely okay with a redesign based on (more) modern components, since a DAC doesn't really have a "distinct" hard-to-emulate nostalgic sound, as the aforementioned music chips do. With an open source DAC redesign, we get to avoid classic bugs and design mistakes that Creative initially made, such as DMA clicking, etc. At the same time, we'd be steering clear of copyright issues.

But if we're going to do an open source SB compatible DAC redesign based on modern hardware anyway, then again, it would be really cool if we could base it on the RISC-V architecture. Let's make this thing completely open and Free. 😀

digger wrote:

I'd like to make a distinction between the "music" part and the DAC part. For the music part, personally I would like to see original hardware used as much as possible, particularly the OPL3 chips and the Tandy compatible sound chips. And we wouldn't have to start from scratch here. We could simply "piggy-back" on other existing open source retro sound card projects such as the Resound OPL3 card and the Tandy 1000 ISA sound card. For implementing the MPU-401 compatibility, perhaps we could (at least in part) adopt and integrate the design from the HardMPU project.

That would leave only the DAC part for us to newly implement. For this, I'm completely okay with a redesign based on (more) modern components, since a DAC doesn't really have a "distinct" hard-to-emulate nostalgic sound, as the aforementioned music chips do. With an open source DAC redesign, we get to avoid classic bugs and design mistakes that Creative initially made, such as DMA clicking, etc. At the same time, we'd be steering clear of copyright issues.

But if we're going to do an open source SB compatible DAC redesign based on modern hardware anyway, then again, it would be really cool if we could base it on the RISC-V architecture. Let's make this thing completely open and Free. 😀

Exactly. The music part is where preserving the feel of older hardware comes into play.

digger wrote:

For the music part, personally I would like to see original hardware used as much as possible, particularly the OPL3 chips and the Tandy compatible sound chips.

I already have mentioned before, that with this approach everything is WAY easier

digger wrote:

And we wouldn't have to start from scratch here. We could simply "piggy-back" on other existing open source retro sound card projects such as the Resound OPL3 card and the Tandy 1000 ISA sound card.

Probably, not required - the logic behind is pretty simple... For example, my C/MS card - is one evening project for schematics, PCB layout and VHDL code, and another hour - for soldering/testing.
Little issue - mixer/configuration control etc

digger wrote:

For implementing the MPU-401 compatibility, perhaps we could (at least in part) adopt and integrate the design from the HardMPU project.

I believe in my the very first post in this thread I suggested using HardMPU 😀 Even more - while waiting for better kiting conditions 🙁 , I have looked into HardMPU code and probably know how to port it to my universal card. Will need to make an optocoupler interface and test it.

digger wrote:

But if we're going to do an open source SB compatible DAC redesign based on modern hardware anyway, then again, it would be really cool if we could base it on the RISC-V architecture. Let's make this thing completely open and Free. 😀

I don't understand it... You can take Altera (now Intel) FPGA, download free version of IDE and use it for the project. How is it different from that RISC-V in a practical way ???

newold86 wrote:

I don't understand it... You can take Altera (now Intel) FPGA, download free version of IDE and use it for the project. How is it different from that RISC-V in a practical way ???

Not much considering RISC-V isn't substantially available outside of FPGA form.

I wouldn't have any issue with a card using a microcontroller port of, for example, Nuke OPL so long as it is essentially indistinguishable from a genuine Yamaha chip. For that matter, it's not impossible that substantial amounts of compatibility logic could be ported across from DOSBox. It all comes down to timing and processing power requirements.

newold86 wrote:

I don't understand it... You can take Altera (now Intel) FPGA, download free version of IDE and use it for the project. How is it different from that RISC-V in a practical way ???

My apologies. I remember someone in this topic mentioning a microcontroller-based solution (specifically an ARM-based one) and someone else (probably you) suggesting an FPGA-based solution. In the case of a microcontroller-based DAC, I think it would be cool if we could use a RISC-V-based microcontroller, such as the aforementioned Freedom E310. I would prefer that over ARM, because of the openness, as well as the coolness of applying a shiny new CPU architecture to a retro project. But since you are apparently pretty far along towards completing a working FPGA-based DAC design, I won't stand in the way of the most obvious and practical solution. 😀

I would be fine either way (FPGA or microcontroller) for the DAC, for the mixer and if practical also for the HardMPU, again, as long as genuine chips are used for the music synthesizer parts. But I believe you and I and most of the people here are on the same page in that regard, right?

digger wrote:

I would prefer that over ARM, because of the openness,

OK, I will repeat my question again 😀 - what does "openness" mean in this situation ? Just trying to understand how this RISC-V is different from, for example, STM32 ARM MCU in our particular case ? Would you mind to provide some example why it could be important from practical point of view ?

digger wrote:

But since you are apparently pretty far along towards completing a working FPGA-based DAC design, I won't stand in the way of the most obvious and practical solution. 😀

I think most efficient way (someone has already mentioned it) is to use relatively small FPGA and separate MCU. For now I'm putting everything into FPGA because it's most flexible way to experiment... After everything is working, the final design could be easily changed to FPGA/MCU

digger wrote:

I would be fine either way (FPGA or microcontroller) for the DAC, for the mixer and if practical also for the HardMPU, again, as long as genuine chips are used for the music synthesizer parts. But I believe you and I and most of the people here are on the same page in that regard, right?

One part of me wants to do everything in FPGA, but it will increase the complexity of the project to unrealistic (at least, to me) level. In any case, after everything is working with real music chips, it's absolutely no problem to replace a music chip with FPGA code (if someone makes it)

newold86 wrote:

OK, I will repeat my question again 😀 - what does "openness" mean in this situation ? Just trying to understand how this RISC-V is different from, for example, STM32 ARM MCU in our particular case ? Would you mind to provide some example why it could be important from practical point of view ?

If you were consider merely the development of the firmware for the microcontroller, you'd be right. One could develop a completely free firmware either way. But this proposal encompasses not just a new firmware, but a new hardware design for a retro sound card. And it just seems like a nice opportunity to both promote and gain experience with a completely open and royalty-free processor architecture. You're right. From a practical standpoint, there wouldn't be an immediate benefit, but I like the idea of an open industry-standard royalty-free CPU architecture and personally I'd jump at any opportunity to encourage its use and growth. But if this proves too impractical compared to a more traditional ARM-based microcontroller, then I'd absolutely be willing to let the idea go for the sake of such a project being successful. 😀

How about the Zilog 80 ? It's old, had been copied a thousand times and it's an eternal classic. 😀