Early Adlib/SB games don't do proper delays when accessing OPL registers, which often causes problems on fast CPUs.
You can try to disable caches to slow down the CPU.

BTW: your card is properly initialized, and passes all DIAGNOSE tests, right?

Yes card works overall great. Its just the older stuff where you dont have sound setup. No Output.
First i was thinking the card needs to be IRQ7 because its hardcodet in the game. But nothing on IRQ5 or 7.
I will slow down the CPU and try again.

If the CT1600 worked fine, then it's likely not a CPU speed issue.

I also doubt it has anything to do with the age of the games. Try Commander Keen 4 and see if it 1) shows "Sound Blaster/Adlib" as supported sound device on initialization screen and 2) outputs sound during gameplay when Adlib/Sound Blaster is selected in options menu.

Your IRQ and DMA are for digital sound and have absolutely nothing to do with Adlib/OPL2/OPL3 support, which happens either on the SB port (220h) or the legacy Adlib port, which is hardcoded to 388h.

The CT2230 is a jumpered, non-PnP card and thus attempting to "initialize" it with e.g. DIAGNOSE will have no effect. But there are no jumper settings that turn on or off Adlib support anyway.

It also has real OPL3, and thus appears identical to a CT1600 to any software trying to use OPL2/OPL3 support.

If Adlib support doesn't even work in Keen, I would suggest there is something wrong with your card.

keenmaster486 wrote on 2025-08-27, 17:24:

The CT2230 is a jumpered, non-PnP card and thus attempting to "initialize" it with e.g. DIAGNOSE will have no effect. But there are no jumper settings that turn on or off Adlib support anyway.

Earlier variants of SB16 are fully jumper-configured, and don't need any initialization.

CT2230 only has jumpers to select the address, while IRQ and DMA are set by software - DIAGNOSE (or UNISOUND) in AUTOEXEC.BAT is required!

Grzyb wrote on 2025-08-27, 17:59:

CT2230 only has jumpers to select the address, while IRQ and DMA are set by software - DIAGNOSE (or UNISOUND) in AUTOEXEC.BAT is required!

Oh, you're right. Well, in any case, IRQ and DMA has nothing to do with Adlib support.

I pluged in a CT1600 for testing and i got sound. So no CPU issue at all. The CT2230 run DIAGNOSE without errors.

Jackhead wrote on 2025-08-27, 18:43:

I pluged in a CT1600 for testing and i got sound. So no CPU issue at all. The CT2230 run DIAGNOSE without errors.

But have you tried to disable the internal and external cache?

CT1600 has a discrete OPL3 chip
CT2230 had OPL3 integrated in the bus interface chip - CT1747

There can be some differences between them...

Yes i slowed down my system. No L1 or even no CPU cache . Nothing changed. Looks like the Beholder dont like to see the 2230 😀

Currently, you have DIAGNOSE in AUTOEXEC.BAT, right?
You can try replacing it with UNISOUND.

If that fails - I guess some weird incompatibility somewhere...

Just occurred to me that it's possible Beholder does some kind of autodetection that fails on this card because of an initialization issue.

Keep in mind that "soundblaster aware" software writes to the OPL on port 22xh (I forget the exact offset from the sb base port) but adlib software writes to it on port 338h. I.e. the card has to listen on both ports. If SB software plays FM but adlib software does not, it would suggest an address decoding problem for port 0x338h. You could start by checking the ISA connector and slot for corrosion and cleanig them

Ok it was a driver problem. Reinstalled with the 2230 floppy's from the archive here. Pretty sure I was using wrong drivers ..

OPL2/OPL3 software does not use drivers.

I'm beginning to think this was a mixer issue.

keenmaster486 wrote on 2025-08-29, 17:48:

OPL2/OPL3 software does not use drivers.

But the bus interface chip may need software initialization to place the OPL chip at the right I/O address.

The 220h address is jumper-selected, but some software is still necessary to set the IRQ and DMA - and probably to enable the 388h address.