The AZT2320 chip has real OPL3 integrated (some of these chips has the OPL logo on then), akin to the CT1747 chip found on some SB16 and AWE32 variants.

All of these games mentioned barely used the capabilities of the OPL3 chip. At least one that I know that used it very well was Cybersphere (Cybersphere Plus counts too).

Gmlb256 wrote on 2023-01-27, 22:01:

The AZT2320 chip has real OPL3 integrated (some of these chips has the OPL logo on then), akin to the CT1747 chip found on some SB16 and AWE32 variants.

It certainly sounds different enough to consider, that even if licensed, it is different. I'm not talking about license, even a NEC V20 was a licensed 8086 and worked very different.

Gmlb256 wrote on 2023-01-27, 22:01:

All of these games mentioned barely used the capabilities of the OPL3 chip. At least one that I know that used it very well was Cybersphere (Cybersphere Plus counts too).

Interesting, I may add it to the comparisons, thanks for the suggestion.

claunia wrote on 2023-01-27, 21:42:

Creative EMU10K1:

While OPL is emulated via software, it sounds much better, more similar to a real OPL chip. However in CPU intensive games the music gets out of sync. And one of the games in my test could not work, in any mode, with this card (X-COM UFO Defense).

FM synth slowdown on SBLive cards depends on the driver version. Later drivers (including the one which ships with Audigy cards) do not have this problem.

ESS AudioDrive (ESS1868F):

It also has an enhanced mode but on my tests only WarCraft II supported it, and well there, it sounds better than the SB mode (aka OPL).

The native ESFM mode was only supported by newer DOS games which use more recent versions of the Miles Sound System. It can indeed sound pretty good if properly utilized. Some tests and observations can be found here: ESFM details

LS-212:
An unauthorized OPL clone of unknown origin that was very common. So common even Creative used it in cards that you would expect to find the real deal.

It is awful, it sounds numb, stay far away.

The LS-212 is supposedly a pin compatible copy of the YMF262-M according to this article. It should sound exactly like the real thing, unless the sound card where it was used botched something with its implementation. Some of your recordings show unusual stuttering and slowdowns, which indicates the latter.

- Why have you not changed this setting and this other one so it sounds better?
Because the idea is to listen to how they sound by default. Yes I know you can disable the echo chamber in the AWE cards. But not everybody does know every option. So defaults are used.

Out of curiosity, did you make the recordings in-game? Or was it done externally from the config utility and/or by playing back MIDI files through a third-party program? Some games like Duke 3D can sound quite different depending on which method is used.

Joseph_Joestar wrote on 2023-01-28, 04:28:

FM synth slowdown on SBLive cards depends on the driver version. Later drivers (including the one which ships with Audigy cards) do not have this problem.

I used the latest Live named drivers I could find, but as well, I didn't use the fastest DOS compatible machine.

Joseph_Joestar wrote on 2023-01-28, 04:28:

The LS-212 is supposedly a pin compatible copy of the YMF262-M according to this article. It should sound exactly like the real thing, unless the sound card where it was used botched something with its implementation. Some of your recordings show unusual stuttering and slowdowns, which indicates the latter.

Yes it is pin compatible, but well I do not have a "sounds amazing with a YMF let's solder an LS212 in it" donor card.
I checked the traces (no rust) and capacitors (within spec) of the LS212 card so all botching is, unfortunately, by design.

And I thing it is important to have that into account, if you get a card that used a cheap clone instead of an original, expect the rest of the card to be garbage as well.

Joseph_Joestar wrote on 2023-01-28, 04:28:

Out of curiosity, did you make the recordings in-game? Or was it done externally from the config utility and/or by playing back MIDI files through a third-party program? Some games like Duke 3D can sound quite different depending on which method is used.

All recordings are in-game (that's why in Dune II some videos have the voice over and some don't, the SETUP utility gets crazy sometimes, yes even the updated one).

claunia wrote on 2023-01-28, 09:18:

Yes it is pin compatible, but well I do not have a "sounds amazing with a YMF let's solder an LS212 in it" donor card.
I checked the traces (no rust) and capacitors (within spec) of the LS212 card so all botching is, unfortunately, by design.

What card is hosting the LS-212? Can you post a picture?

While I don't have any cards with that exact chip, I do have a few with the FT6116-100 and DXP44Q copies, which are also mentioned in the article that I linked to earlier. The FM synth on those cards sounds exactly the same as on my Yamaha YMF719, and there is no stuttering or slowdown during playback.

And I thing it is important to have that into account, if you get a card that used a cheap clone instead of an original, expect the rest of the card to be garbage as well.

Agreed. Those copy chips were mostly used by "no-name" generic card manufacturers. That said, other than higher self-noise, I haven't had any issues with FM synth music on cards which use the aforementioned copy chips.

What exactly are you testing?

A lot of the comments you make are related to FM synth (OPL3), but you also refer to (MIDI wavetable) sound fonts, and both GM and MT-32. You then list only chips, not sound cards, and some of thost chips can only do one of these options, and for other cards you refer to software (MIDI?) synthesis. You also state that the LS-212, which is known to be a 1:1 knockoff of the YMF-262 sounds different. It doesn't, indeed it can't, any difference is due to the analog parts of the cards you are testing, not the chips themselves.

I appreciate the work that must have gone into this, but as written up here, the comparison seems to miss a lot of rigor to make sure you're comparing the same functionality on different cards, and that when you refer to chips, you are actually describing differences between chips and not just the cards they were stuck on.

i started a separate thread here inquiring about the possible opl3 core inside the aztech azt-2320 chip. i'm now convinced it's either a licensed or unlicensed 1:1 clone, but the sound does have a very slightly different character to my 20+ year cqm-damaged ear. i mean it could be down to filtering or the construction or age of the cards--both cards i'm sampling from are oem builds. someone in yet another thread suggested the opl3 core is from a ymf289, but i've never even heard an opl3-l (or opl3-sax) to discern a difference.

i found this opl3 clone testing tool that might be interesting to try. i don't know how authoritative it is in confirming a 1:1 clone, but it certainly seems definitive if the chip fails the test.

The way to detect OPL3 clone

dionb wrote on 2023-01-28, 16:13:

What exactly are you testing?

A lot of the comments you make are related to FM synth (OPL3), but you also refer to (MIDI wavetable) sound fonts, and both GM and MT-32. You then list only chips, not sound cards, and some of thost chips can only do one of these options, and for other cards you refer to software (MIDI?) synthesis. You also state that the LS-212, which is known to be a 1:1 knockoff of the YMF-262 sounds different. It doesn't, indeed it can't, any difference is due to the analog parts of the cards you are testing, not the chips themselves.

I appreciate the work that must have gone into this, but as written up here, the comparison seems to miss a lot of rigor to make sure you're comparing the same functionality on different cards, and that when you refer to chips, you are actually describing differences between chips and not just the cards they were stuck on.

I'm comparing how it sounds.
I'm not comparing what's the best OPL clone.

I'm comparing which chip generates the best sound.

Different cards can generate different sounds with the same chip?
Yes, and 99.99999% of people could not hear them unless:
a) The card is 10x crappier than the others (most improbable, the cheapest the card the more it will follow whatever the manufacturer datasheet says to make a PCB with, making decisions is engineering time and that's more expensive than tracers or capacitors)
b) The card is damaged (and for your comments I'm starting to think my LS-212 card is damaged even if all tests indicated otherwise, I'm searching another)

This is not a "let's compare the waveform" (spoiler: it's different in all cards), it's a "let me listen to it and pick the one I like the best how it sounds", and that is chip dependent, not card dependent.

If an OPL clone makes a note sound too metallic, it's going to sound too metallic whatever the card that clone is installed to, that's not affected by the rest of the circuitry.

That's the comparison, simple, "which one do I like most how it sounds"

dionb wrote on 2023-01-28, 16:13:

for other cards you refer to software (MIDI?) synthesis.

Uhm, wait what?

Maybe confusion somewhere?

Creative CTxxxx (as well as Ensoniq ESxxxx) and Creative EMU10K1 and higher use software FM synthesis (OPL emulation), as well as the VIA VT8231.
Macintosh uses software MIDI synthesis.

All others do it on hardware, whichever they support.

Joseph_Joestar wrote on 2023-01-28, 04:28:

LS-212:
An unauthorized OPL clone of unknown origin that was very common. So common even Creative used it in cards that you would expect to find the real deal.

It is awful, it sounds numb, stay far away.

The LS-212 is supposedly a pin compatible copy of the YMF262-M according to this article. It should sound exactly like the real thing, unless the sound card where it was used botched something with its implementation. Some of your recordings show unusual stuttering and slowdowns, which indicates the latter.

Having an identical pinout does not imply having identical internals. The LS-212 sounds very different, and frankly, horrible.

jwt27 wrote on 2023-01-30, 01:59:

Having an identical pinout does not imply having identical internals. The LS-212 sounds very different, and frankly, horrible.

Fair enough, we don't know what's inside the copy chip. As mentioned before, I don't have a card with an LS-212 so I can't really say much about that. But the fact that there are slowdowns and stuttering in the recordings makes me think that something else might be going on with the card hosting that chip.

I can however offer a comparison between the YMF262-M and its (supposedly) 1:1 pin compatible copy the FT6116-100. The original chip is on a Vibra16 sound card while the copy is on an Avance Logic ALS100. I didn't do a full waveform comparison, so I can't claim that these are 100% identical, but they sound fairly similar to my ears. Aside from the ALS100 card having more self-noise of course, and the volume levels being slightly different.

Joseph_Joestar wrote on 2023-01-30, 03:46:

jwt27 wrote on 2023-01-30, 01:59:

Having an identical pinout does not imply having identical internals. The LS-212 sounds very different, and frankly, horrible.

Fair enough, we don't know what's inside the copy chip. As mentioned before, I don't have a card with an LS-212 so I can't really say much about that. But the fact that there are slowdowns and stuttering in the recordings makes me think that something else might be going on with the card hosting that chip.

I can however offer a comparison between the YMF262-M and its (supposedly) 1:1 pin compatible copy the FT6116-100. The original chip is on a Vibra16 sound card while the copy is on an Avance Logic ALS100. I didn't do a full waveform comparison, so I can't claim that these are 100% identical, but they sound fairly similar to my ears. Aside from the ALS100 card having more self-noise of course, and the volume levels being slightly different.

Waveforms are going to be different, even between same model cards, as analogue components are never 100% identical, not even when new.

Question is WHAT is different.

If the waveform is lower but has the exact same vectors, it's a volume difference. I have normalized the volume on all videos to prevent this, even the LS-212 that still sounds numb.

But filtering (and bad capacitors) can cut a waveform flat, so there is no peak. Yet the rest of the waveform is identical. This is not really perceptible by the human ear in the middle of a song unless it's ALL notes or VERY flattened (and this is what I referred with 10x crappier or damaged card).
You can check if you compare the ESS1868F with the ES1946S videos. The waveform is different, but the notes are the exact same, just my ESS1868F is much older, needs more capacitors that are older, and that means the signal is noisier. But if you notice anything, it's noise, not a different music.

However if in the same exact note the waveform is radically different, that's not the card, it's the chip, that's not doing an identical synthesis.
And this is easily perceptible just by ear in most cases. (And made my life difficult to align the songs in many cases, as except the AZT, CMI, YMF, DaVinci was unable to auto align the rest of recordings!!!)

jwt27 wrote on 2023-01-30, 01:59:

Having an identical pinout does not imply having identical internals. The LS-212 sounds very different, and frankly, horrible.

Also, everybody talks about the LS-212, but forget about the LS-215.

The YMF chip generates digital, floating point, sound, that needs to be converted into analogue sound by a DAC, usually the YAC512.

The LS-212 is pin compatible with the YMF262, and the LS-215 is pin compatible with the YAC512.

And combinations have been found, with a Yamaha DAC with the clone LS synth. And viceversa.
This certainly affects the final waveform.

My card is a LS-212 with LS-215 DAC.

claunia wrote on 2023-01-30, 08:46:

But filtering (and bad capacitors) can cut a waveform flat, so there is no peak. Yet the rest of the waveform is identical. This is not really perceptible by the human ear in the middle of a song unless it's ALL notes or VERY flattened (and this is what I referred with 10x crappier or damaged card).

My bad, I was watching this video and thought that the slowdowns and stuttering came from the LS-212, when they were actually produced by the VT8231.

The LS-212 sounds as if it's playing underwater, i.e. like a very strong low pass filter is being erroneously applied to the output. I'm still curious as to what card exactly is hosting the chip?

Joseph_Joestar wrote on 2023-01-30, 10:04:

claunia wrote on 2023-01-30, 08:46:

But filtering (and bad capacitors) can cut a waveform flat, so there is no peak. Yet the rest of the waveform is identical. This is not really perceptible by the human ear in the middle of a song unless it's ALL notes or VERY flattened (and this is what I referred with 10x crappier or damaged card).

My bad, I was watching this video and thought that the slowdowns and stuttering came from the LS-212, when they were actually produced by the VT8231.

The LS-212 sounds as if it's playing underwater, i.e. like a very strong low pass filter is being erroneously applied to the output. I'm still curious as to what card exactly is hosting the chip?

VT8231 FM synthesis is in software.

The card model is in the video description, it's a Zoltrix 6400