Since, AFAIK, Creative has not made the AWE64 chip spec public, how is it even theoretically possible to do anything other than reusing existing CT8903/CT8920 chips "as is" (which means, removing them from the original PCB)?

Dump it, reverse engineer it etc. Obviously it would be a much bigger and costlier project but it's not 'impossible' considering the kind of stuff people have been able to do with FPGAs in the past. The upside is that you could build infinite amounts of them, iron out hardware bugs and further improve on the design.

Sorry to hear that, appiah4.
On the other hand, as a Russian saying goes, it's not a cognac to please everyone.
This specific issue has already been discussed on the first page with several fellow members commenting on it. We don't see any problem in "destroying" (as you put it, but I would name it otherwise) several common CT4520's. Moreover, I don't feel any specific love for FPGA projects, nor do we have any experience/funds/drive to make it happen in that form. Sorry that our project hasn't met your expectations.

hard1k wrote:

Sorry to hear that, appiah4.
On the other hand, as a Russian saying goes, it's not a cognac to please everyone.
This specific issue has already been discussed on the first page with several fellow members commenting on it. We don't see any problem in "destroying" (as you put it, but I would name it otherwise) several common CT4520's. Moreover, I don't feel any specific love for FPGA projects, nor do we have any experience/funds/drive to make it happen in that form. Sorry that our project hasn't met your expectations.

It's ok, no need to apologize to me over a bunch of hardware. I will be watching this project closely regardless, and I wish you all the luck with it. You seem like a very nice chap, I hope you make the card of your dreams. 😎

Wow, interesting project to make "perfect" AWE64. 😲
I like the looks of the prototype and WaveBlaster header and "real" OPL3.
But I am not a fan of RAM mounted on the back side of the board. Those ICs are quite thick and can interfere with a card in adjacent slot. Why not place those RAM chips on the front side? 😕
And I don't like SMD electrolytic capacitors.

Destroying old AWE64's is... somewhat not pleasant. But it you'll make a better one from it - I have no reason to complain.
AWE64 Value is pretty common and doesn't represent any unique good features for me. AWE32 is better if you ask me. So if I can get another AWE64 Value, rip chip out and put it onto new board, which will make it better - I'm all for it. And I'll definitely pick one.

hard1k wrote:

We've used the MIDI-OUT port from the gameport to run the waveblaster header. I don't think it would run two devices connected in parallel, but correct me if I'm wrong here.

As far as I know, you need only to connect MIDI OUT to the synth, you don't need MIDI IN. And you can connect both WaveBlaster and external port to the same source (so those just will play the same stuff). Of course, external port will require additional buffering and protection, but it is pretty simple. I'm doing it in my FMonster.

Some comments on questionnaire.

Are you interested in replacing the 1 MiB WaveROM?

As far as I know, you can download any soundbank in that RAM that is already installed to "replace" stock ROM. So why bother with replacing ROM? 😕

Mounting holes for WB

I think one, opposite from the connector is enough. But, probably, we can find a standoff that doesn't require a hole in a board? Like, glued on top of the board with sticky tape? I think it would be easier to find some room for three standoffs of that type to mount WaveBlaster-compatible synth. It is much easier to free part of the board from components than from traces.

And yes, I'm in, I want one.
I'll get kit or bare board if there will be such an option.

Hi there! We've been waiting for your comments, I was starting to get a bit nervous not seeing it after a week since we went public on this 😁

Fagear wrote:

But I am not a fan of RAM mounted on the back side of the board. Those ICs are quite thick and can interfere with a card in adjacent slot. Why not place those RAM chips on the front side? 😕

We had that dispute when the design was still at very early phase. We've opted for placing them on both sides in order to keep the PCB within the ISA Half-Size measures, otherwise it would exceed them. And I know several fellows who are keen of small PCs (industrial / homemade / thin clients, etc.) and even use PC/104 soundcards. Of course, we won't design it for PC/104, but keeping it Half-Size makes perfect sense for me.

Fagear wrote:

And I don't like SMD electrolytic capacitors.

It's still a prototype, the final product might get better solutions. Consider this one as a successful proof of concept, not more, not less.

Fagear wrote:

As far as I know, you need only to connect MIDI OUT to the synth, you don't need MIDI IN. And you can connect both WaveBlaster and external port to the same source (so those just will play the same stuff). Of course, external port will require additional buffering and protection, but it is pretty simple. I'm doing it in my FMonster.

Great, thanks!

Fagear wrote:

As far as I know, you can download any soundbank in that RAM that is already installed to "replace" stock ROM. So why bother with replacing ROM? 😕

I am not *that* experienced in AWE64, but IIRC there were some use-cases when the software couldn't run under Windows and it didn't support AWE in its native mode, so you had to run it with AWEUTIL in some emulation mode, making it use the ROM without any possibility for RAM-based soundfonts. That's where the limitations of ROM became somewhat evident (once again, our survey proves that most people are happy with the original ROM bank, so I shouldn't offend and insult Creative on this one), something similar happened to GUS PnP in similar circumstances. And as ARGUS project proved it viable to replace the ROM with a larger one, and people asked about this option for our AWE64 project, I've decided to include this point in the survey. As I've mentioned, so far we seem to stick to the original one.

Fagear wrote:

I think one, opposite from the connector is enough. But, probably, we can find a standoff that doesn't require a hole in a board? Like, glued on top of the board with sticky tape? I think it would be easier to find some room for three standoffs of that type to mount WaveBlaster-compatible synth. It is much easier to free part of the board from components than from traces.

Actually the main argument against additional holes consists in that they would presumably make the overall design (in aesthetic terms) worse. Not that we can't free the board from traces or components - in fact we can. We just want to know how many people actually need it. If there is some critical mass in order to convince those who want to sacrifice the holes for the sake of aesthetics, we'll make those holes.

Neat project. I'm interested in picking one up -- depending on the final price.

dr_st wrote:

Since, AFAIK, Creative has not made the AWE64 chip spec public, how is it even theoretically possible to do anything other than reusing existing CT8903/CT8920 chips "as is" (which means, removing them from the original PCB)?

There might be some unused chips available from various surplus component stores. Not sure about CT8903/CT8920 though.

https://www.win-source.net/unclassified-emu8000.html
https://www.win-source.net/unclassified-emu8011-01.html

hard1k wrote:

We've been waiting for your comments, I was starting to get a bit nervous not seeing it after a week since we went public on this 😁

A was too busy at work, sorry. 😵

hard1k wrote:

We've opted for placing them on both sides in order to keep the PCB within the ISA Half-Size measures, otherwise it would exceed them.

Oh, I see. I understand the reasoning.
But in my opinion there is plenty of space left on the board free of components. A small shift of OPL3 components to the right, subsequent shift of main IC and there will be space for 2nd row for RAM ICs where the logo is now. (But that's just me, I really don't like components on the rear side of PCBs. 😀 )

hard1k wrote:

It's still a prototype, the final product might get better solutions.

I get it. Just keep in mind, that those SMD caps are harder to solder than TH ones. Those take up more space due to plastic "skirt" and bent out leads than TH ones. And SMD electrolytics are prone to leak.

hard1k wrote:

I am not *that* experienced in AWE64, but IIRC there were some use-cases when the software couldn't run under Windows and it didn't support AWE in its native mode, so you had to run it with AWEUTIL in some emulation mode, making it use the ROM without any possibility for RAM-based soundfonts. That's where the limitations of ROM became somewhat evident (once again, our survey proves that most people are happy with the original ROM bank, so I shouldn't offend and insult Creative on this one), something similar happened to GUS PnP in similar circumstances

I'm not too experienced with AWE64 either, so it seems like there is a good reason to make ROM swappable.

ZipoBibrok wrote:

There might be some unused chips available from various surplus component stores. Not sure about CT8903/CT8920 though. […]
Show full quote

dr_st wrote:

Since, AFAIK, Creative has not made the AWE64 chip spec public, how is it even theoretically possible to do anything other than reusing existing CT8903/CT8920 chips "as is" (which means, removing them from the original PCB)?

There might be some unused chips available from various surplus component stores. Not sure about CT8903/CT8920 though.

https://www.win-source.net/unclassified-emu8000.html
https://www.win-source.net/unclassified-emu8011-01.html

That's interesting and is worth checking.

Cheapest options would be to opt for 72pin SIMM slot. If one of those RAM IC's die it's a pain in the ass to replace one.

Unfortunately that's exactly one of the features that make this project different, so we won't go for a RAM slot.

Moreover, making it cheap was never among the project objectives. Those who want a cheap AWE can easily get a regular original CT4520 for $30 at most, there are plenty of them.

Well, from my point of view, I would ditch OPL for now and concentrate in what is easy . Like SPDIF, WT header and integrated ram.

OPL3 is just another key feature, people are very excited about it, and its absence was the major drawback if the original card. Unless we absolutely fail in making it work as supposed, we won't drop it.

hard1k wrote:

OPL3 is just another key feature, people are very excited about it, and its absence was the major drawback if the original card. Unless we absolutely fail in making it work as supposed, we won't drop it.

Agreed; it is one of the crowning features of the design. If they drop that, is there really a reason to do it at all?

carlostex wrote:

Cheapest options would be to opt for 72pin SIMM slot. If one of those RAM IC's die it's a pain in the ass to replace one.

Unless you get into low-end parts of questionable reliability, these sorts of sockets and headers are almost never cheap. For someone who doesn't already have the appropriate SIMM, this will likely be cheaper in the long run.

Real OPL3 is for purists, most people won't even notice the diference. AWE64 was not a card made for MSDOS, but for windows. So CQM is just fine, MIDI is the keypoint, and SPDIF an extra.
I didn't say to remove OPL3, I just said, for now, it's better to concentrate on other stuff, CQM is there for now to emulate OPL3.

I would consider designing the AWE64 legacy to use the backplates from the AWE64 values you are stealing the chipset from. Unless you already have an affordable solution for the back plate of course.

The RCA ports are really not needed, you can get far more than enough audio quality out of a 3.5mm Audio connector.

Regarding OPL3, I would consider that one of the premier goals of this card, so removing it would be an absolute last resort thing,

Marmes wrote:

Real OPL3 is for purists, most people won't even notice the diference. AWE64 was not a card made for MSDOS, but for windows. So CQM is just fine, MIDI is the keypoint, and SPDIF an extra.
I didn't say to remove OPL3, I just said, for now, it's better to concentrate on other stuff, CQM is there for now to emulate OPL3.

I'm pretty sure that you would be in the extreme minority of people interested in this card with that opinion.

Lack of genuine OPL is the number one reason people are dissatisfied with the stock AWE64 (and relevant AWE32 models)

Then why choose this AWE64 chip? Why not AWE32? Or other. There is no information about this chip. So unless you know how to disable CQM correctly, there is no point in implementing something that will not work correctly. So what I proposed was to ditch OPL3 for now and later after some research , implement again. Being a minority, I wouldn't mind having a card like AWE64 with all the ram for sound, WT header and SPDIF out. Isn't that a good improvement while there is no OPL3 runing yet?

maxtherabbit wrote:

Marmes wrote:

Real OPL3 is for purists, most people won't even notice the diference. AWE64 was not a card made for MSDOS, but for windows. So CQM is just fine, MIDI is the keypoint, and SPDIF an extra.
I didn't say to remove OPL3, I just said, for now, it's better to concentrate on other stuff, CQM is there for now to emulate OPL3.

I'm pretty sure that you would be in the extreme minority of people interested in this card with that opinion.

Lack of genuine OPL is the number one reason people are dissatisfied with the stock AWE64 (and relevant AWE32 models)

Marmes wrote:

Then why choose this AWE64 chip? Why not AWE32? Or other.

The AWE32 has multiple chips, versus just one in the AWE64, so integration of the AWE64 is easier. Furthermore, the DSP AWE64 has fewer bugs.

Marmes wrote:

I wouldn't mind having a card like AWE64 with all the ram for sound, WT header and SPDIF out.

So I ask you - why not get an AWE32? There are versions out there with WT header, standard RAM slots, and even OPL3. Why would you care about the AWE64 Legacy? Just for SPDIF?