RayeR wrote on 2022-11-03, 12:02:

Ok, but why they degraded the quality on newer cards? I would understand if they replaced 3rd party chip with theirs cheaper one but this OPL is already from Creative (at least by marking)...

The OPL part is a licensed Yamaha design integrated into the Creative chip. Creative Labs designed a music synthesis solution that sounds similar enough for a casual listener, but is different enough to avoid infringing "intellectual property" by Yamaha. The goal was to get rid of license fee, not to get rid of Yamaha chips.

Aha, this makes sense... But what about other chip manufacturers like crystal and ess? Did they pay for licenses even are cheap clones? Or are also different enough but better quality?

Other companies like Crystal and ESS didn't sell the cards, only chips to people who make the cards, and to these people it was probably more of a cost thing. Not needing one more expensive chip helps with profits and makes your chip more attractive.
Unlike Creative, ESS and Crystal actually knew what sound was and how it works (Both are leading sound related chips makers to this day) and could develop their own flavor that worked much better. Some people like Crystal and ESS even more than Yamaha's original, and partly because it sounds right for much more, if not all things, unlike CQM which can sound pretty broken.

RayeR wrote on 2022-11-03, 12:53:

Aha, this makes sense... But what about other chip manufacturers like crystal and ess? Did they pay for licenses even are cheap clones? Or are also different enough but better quality?

Just adding to Tiido's reply about the quality: Yes, both ESS and Crystal also developed their license-free in-house OPL3 substitutes. ESS actually built a more complex FM synthesizer that can be switched into either its native mode or into OPL3 compatible mode. Their flavour of FM synthesis is called "ESFM". AFAIK the Windows 95 MIDI driver actually uses the extended capabilities of ESFM. Another Vendor that has an in-house OPL3 substitute is Analog Devices. While they provided the "reference" for the "Windows Sound System" (which is approximately defined as: this is an AD1848 on an ISA card with minimal glue logic), sound cards using an AD1848 (or similar chips by Crystal) had external solutions for SoundBlaster compatibility. They were late to the game with their own SoundBlaster compatible chip, the AD1816, which is not AD1848 compatible. This is unlike many earlier cheap SB/WSS cards that could be switched into a WSS compatible mode. The AD1816 datasheet is extremely misleading about WSS compatibility. The AD1816 fulfills all hardware requirements to be qualified as "sufficiently good sound card for Windows", which basically means 48kHz stereo sound at 16 bits for both recording and playback. So a driver integrating into Windows 95 can be made that provides "full fidelity". This appears to be enough for the data sheet to claim "supporting the Windows Sound System". The AD1816 OPL3 substitute is not highly regarded in the retro community, although I am not sure there is a consensus whether ADs OPL3 substitute or CQM is worse.

Tiido wrote on 2022-11-03, 13:31:

Unlike Creative, ESS and Crystal actually knew what sound was and how it works (Both are leading sound related chips makers to this day) and could develop their own flavor that worked much better.

Subjectivity aside, it was Dave Rossum, of Creative subsidiary E-mu Systems, who developed CQM. He knows a thing or two about sound and synthesis. 😉

CQM's biggest fault seems to be its name. Had it been called "EQM," it would likely be viewed in the same light as the other FM derivatives.

Sorry to disrespect the guy, but the result seems not to be favorable at all, at least in my limited experience. (I am thinking of making a small batch of CQM sound cards, using the chips I have gathered with all the de-CQMs I have done, just for the variety 🤣)

Cloudschatze wrote on 2022-11-03, 15:32:

Tiido wrote on 2022-11-03, 13:31:

Unlike Creative, ESS and Crystal actually knew what sound was and how it works (Both are leading sound related chips makers to this day) and could develop their own flavor that worked much better.

Subjectivity aside, it was Dave Rossum, of Creative subsidiary E-mu Systems, who developed CQM. He knows a thing or two about sound and synthesis. 😉

CQM's biggest fault seems to be its name. Had it been called "EQM," it would likely be viewed in the same light as the other FM derivatives.

it sounds bad simple as

in the immortal words of the rock "it doesn't matter what your name is"

It's certainly subjective. I'm personally not a fan of the the high-frequency, "aliasing" noise present in some CQM examples, but find the output to be acceptable for most things. At the same time, CQM and the other non-Yamaha implementations are FM derivatives; I've never thought it reasonable or necessary to expect 1:1 OPL3 exactness with any of them.

Cloudschatze wrote on 2022-11-03, 16:46:

It's certainly subjective. I'm personally not a fan of the the high-frequency, "aliasing" noise present in some CQM examples, but find the output to be acceptable for most things. At the same time, CQM and the other non-Yamaha implementations are FM derivatives; I've never thought it reasonable or necessary to expect 1:1 OPL3 exactness with any of them.

I don't either, I greatly enjoy ESFM and CrystalFM. CQM just sounds bad it's really that simple

Tiido wrote on 2022-11-03, 13:31:

Other companies like Crystal and ESS didn't sell the cards, only chips to people who make the cards, and to these people it was probably more of a cost thing. Not needing one more expensive chip helps with profits and makes your chip more attractive.
Unlike Creative, ESS and Crystal actually knew what sound was and how it works (Both are leading sound related chips makers to this day) and could develop their own flavor that worked much better. Some people like Crystal and ESS even more than Yamaha's original, and partly because it sounds right for much more, if not all things, unlike CQM which can sound pretty broken.

I am in the Crystal > ESS = OPL > CQM camp, for example.

If you can find a premium ISA card with a Crystal or ESS chip, then you will end up with magnitudally better sound than any Creative card..

I see those chips can be bought. If values for the passives are known it can be done.

I've got a CT3600 with CQM. I'm certainly capable of the soldering, but that card just never shined for me. My understanding is that it has to do with TSR and protected mode games, being useless unless implemented in the game itself.

r00tb33r wrote on 2022-11-05, 06:19:

I've got a CT3600 with CQM. I'm certainly capable of the soldering, but that card just never shined for me. My understanding is that it has to do with TSR and protected mode games, being useless unless implemented in the game itself.

Well, you need the TSR (AWEUTIL) only for one specific use case: Getting wave-table synthesized music out of a game that outputs MIDI music. The idea by Creative Labs was that hardware implementation of a full MIDI parser is too expensive and complicated, so they shifted that job to software. Modern software was supposed to directly implement support for the AWE32 - and most later games in fact contain the software required to generate AWE32 programming values from MIDI commands. Many earlier games with General MIDI music were not using protected mode, so there is just a small time frame of games that run in protected mode but don't have native AWE32 support.

There is another use case considered relevant today that Creative Labs didn't care about: The "native AWE32 support" included with games (using code/data provided by Creative Labs) always use the ROM sound samples of the AWE32 card. The ROM is just 512KB in size. As long as the memory sockets are not broken, you can add 2MB of sample RAM to your card at quite low cost, and load superior sample sets (called "sound fonts" by Creative Labs) into the RAM. Games are not going to use them, though. On the other hand, the AWEUTIL TSR is able to use all kind of sound fonts, whether they are referencing the ROM samples or contain custom samples that need to be loaded into RAM. Using AWEUTIL-based MIDI synthesis thus can sound better than the native AWE32 support of modern games. If that is a use case interesting to you, you might want to try DOS32AWE, a modified DOS extender that allows passing MIDI data from protected mode games to AWEUTIL. DOS32AWE is compatible with most protected mode games.

You do not need any TSR and get perfect support for Sound Blaster 16 digital audio and the FM music interface. After de-CQMing that card, we also can call the FM music "perfect".

I'm currently looking to de-CQM a few cards, just seems like the YMF289's are becoming less available.
Sellers now want $30-40 per chip, making this far less appealing to do vs. buying a different SB16 with true OPL.

If I had to guess, the big new sound cards coming out that use it increased the perceived demand for these chips, so the sellers are getting greedy and throwing the good ol' pandemic shortage excuse to justify it.
I found them somewhere for 21,50 euros that seemingly have now shipped out, which I can still justify (one card it's going on, a CT2940, is a freebie from a colleague of mine, the other card, a CT2950, was cheap enough and has a bad CQM and/or DAC, and I want to upgrade it if I'm already at it), but it's sad that this seems to become less attainable to do.

Same for those AMD Interwaves, stocks are phasing out fast.
Good that PicoGUS is a good solution there, I don't consider that a lesser option than an Interwave (neither are "true" GUSes, from a certain point of view).

Wow they really have become pretty much unavailable 🙁
It's not like the YMF262 could be substituted either, could it? Different operating frequency from what I recall.

I wonder if any modern microcontrollers could emulate the OPL3, would that be better than just running a software emulator?

Thermalwrong wrote on 2023-07-05, 19:10:

Wow they really have become pretty much unavailable 🙁
It's not like the YMF262 could be substituted either, could it? Different operating frequency from what I recall.

I wonder if any modern microcontrollers could emulate the OPL3, would that be better than just running a software emulator?

Not without some sort of "stick-on" PCB with its own YMF262/YAC512 combo, and supporting circuitry (clock crystal and caps mostly), the way the YMF289 interfaces to the YAC516 is quite different from the YMF262/YAC512 combo from what I'm reading in the datasheets.
Said PCB would have to take the data signals to the YMF289 (also present on the CT1978 CQM) and VCC/VSS, and have the output of the YAC wired to the board.

Possible, but unlikely to be a clean solution.

Alternatively, if ISA slots are not a concern, one could possibly get a ReSound OPL3 card to run alongside with a YMF262, and just yank the CQM chip and its DAC off the SB card in question. Some creative (no pun intended) skullduggery would be required to reroute the ReSound's output to where it goes on the ViBRA card, so that the mixer controls continue working etc.

OwO! What a nice tread! I really love this kind of modification. It's really a pity my CT2950 not working yet. But I'm happy that there's such posts like this one and people are really able to make these modifications with success. And as I suspected, the "X3" oscillator component is present when it comes with an OPL3. Seems like an originally manufactured CT2950 with OPL3 is pretty hard to find. Will try it when I got an OPL3.