Truth is I don't care who is thinking such or different and why. Truth is I need more info about CS4237. I am very concerned about this question. You don't want to say. OK.

you can find nice info about CS4237 in this ftp://ftp.alsa-project.org/pub/manuals/cirrus/4237b.pdf. For outputting SPDIF this can be done:
Orange = output , Yellow = Input
the good way

And the not so good way

They go to the DSP Audio data SP. 2nd option can cause a lot of noise because it goes from the analogue mixer, so it outputs all the noise, you can do it in windows. Also things depend on SPDIF receiver, maybe yours is too sensitive. Also Malibu lacks of spdif buffer. But this shouldn't be an issue. At least it's not with my receivers.

appiah4 wrote on 2021-01-05, 14:05:

I'm currently trying out a 8330 (Not sure what revision, never lifted the sticker to check but the 3D spatializer appears to be […]
Show full quote

I'm currently trying out a 8330 (Not sure what revision, never lifted the sticker to check but the 3D spatializer appears to be discrete so it's likely an A) in my Celeron 533 tinkerbox rig (which I use to basically tinker with various AGP/PCI/ISA cards for fun..)

This computer normally has an AWE64 Value in it. I am so pleasantly surprised by this card that I may possibly replace the AWE64 with this card + Dreamblaster S2.

It has decent line out, SPDIF In/Out, Wavetable header, bugfree MPU-401, what seems to be an integrated 100% OPL3 clone and both SB Pro and SB 16 compatibility. It is in theory the perfect ISA sound card. I just wish it had cleaner PCM. But it will do until the day comes when I can get an Orpheus 😁

The DOS drivers it comes with are a bit of a pain in the ass though. I believe UNISOUND covers this card so that's at least some relief.

I thought these chips didn't have SB16 compatibility (only some ALS cards) and that the OPL3 was bad? Interesting.

Oetker wrote on 2021-01-08, 07:11:

I thought these chips didn't have SB16 compatibility (only some ALS cards) and that the OPL3 was bad? Interesting.

The feature set of the CMI8330 is quite interesting. Gerwin did a nice a review of it in this thread: Re: Sound Blaster 16 Clones

Some FM synth recordings can also be found on YouTube.

PARUS wrote on 2021-01-07, 21:09:

It does not matter. Both TTL and COAX are full-fledged proper S/PDIF. They differ in signal amplitude only. If you have TTL (2-3 […]
Show full quote

appiah4 wrote on 2021-01-06, 12:43:

Well, as far as I know its SPDIF In/Out is 5V and not directly compatible with most audio gear, no?

It does not matter. Both TTL and COAX are full-fledged proper S/PDIF. They differ in signal amplitude only. If you have TTL (2-3+ V, often 4-5V) and need COAX you can just use this:

Not sure I get it.. Is CMI8830's SPDIF In/Out TTL or COAX?

The chip itself only outputs TTL, to get coax some extra stuff is necessary but this is true with pretty much all the different chips out there. Coax is 1Vpp into 75ohm load, and usually is transformer coupled to prevent ground loops from being a problem. TTL signals are meant for internal connectivity and they are 3.3 or 5Vpp signals not meant for driving 75ohm loads but conversion to/from is simple. The bracket shown on previous page does a very quick and dirty conversion using only a resistor divider but relies on whatever drives it having high enough current output for it to work (which shouldn't be a problem with 99% things). No transformer though. TOSLINK modules just take TTL signal directly and get galvanic isolation due to optic nature, so no transformers are necessary.

Tiido wrote on 2021-01-08, 09:46:

The chip itself only outputs TTL, to get coax some extra stuff is necessary but this is true with pretty much all the different chips out there. Coax is 1Vpp into 75ohm load, and usually is transformer coupled to prevent ground loops from being a problem. TTL signals are meant for internal connectivity and they are 3.3 or 5Vpp signals not meant for driving 75ohm loads but conversion to/from is simple. The bracket shown on previous page does a very quick and dirty conversion using only a resistor divider but relies on whatever drives it having high enough current output for it to work (which shouldn't be a problem with 99% things). No transformer though. TOSLINK modules just take TTL signal directly and get galvanic isolation due to optic nature, so no transformers are necessary.

Does this mean I can't just add a COAX out bracket and route to my amp? But I can route it internally to say a PCI sound card with optical out?

Yes, direct connection is not gonna work and in worst case will destroy the signal on the card. There needs to be some extra bits inline with the signal, such as two resistors (and ideally a transformer) as seen on the photo in previous page. You can do direct connection from one card's TTL output to another card's TTL input.

Tiido wrote on 2021-01-08, 13:15:

Yes, direct connection is not gonna work and in worst case will destroy the signal on the card. There needs to be some extra bits inline with the signal, such as two resistors (and ideally a transformer) as seen on the photo in previous page. You can do direct connection from one card's TTL output to another card's TTL input.

Ah, I see. So in order to get digital DOS sound I should pair this with a PCI card with optical out that I can initialize without a TSR in DOS.. I guess something like a Terratec DMX Xfire 1024 (CS4624) should work well as a pair? Easy to initialize without a TSR and would probably make quite a decent primary sound card for Windows 2K/XP.

I would simply get a suitable bracket or make my own. You only need one 374ohm resistor and one 90.9ohm resistor in addition to wires and connectors.

appiah4 wrote on 2021-01-08, 09:38:

PARUS wrote on 2021-01-07, 21:09:

It does not matter. Both TTL and COAX are full-fledged proper S/PDIF. They differ in signal amplitude only. If you have TTL (2-3 […]
Show full quote

appiah4 wrote on 2021-01-06, 12:43:

Well, as far as I know its SPDIF In/Out is 5V and not directly compatible with most audio gear, no?

It does not matter. Both TTL and COAX are full-fledged proper S/PDIF. They differ in signal amplitude only. If you have TTL (2-3+ V, often 4-5V) and need COAX you can just use this:

Not sure I get it.. Is CMI8830's SPDIF In/Out TTL or COAX?

It's TTL.

Oetker wrote on 2021-01-08, 07:11:

appiah4 wrote on 2021-01-05, 14:05:

I'm currently trying out a 8330 (Not sure what revision, never lifted the sticker to check but the 3D spatializer appears to be […]
Show full quote

I'm currently trying out a 8330 (Not sure what revision, never lifted the sticker to check but the 3D spatializer appears to be discrete so it's likely an A) in my Celeron 533 tinkerbox rig (which I use to basically tinker with various AGP/PCI/ISA cards for fun..)

This computer normally has an AWE64 Value in it. I am so pleasantly surprised by this card that I may possibly replace the AWE64 with this card + Dreamblaster S2.

It has decent line out, SPDIF In/Out, Wavetable header, bugfree MPU-401, what seems to be an integrated 100% OPL3 clone and both SB Pro and SB 16 compatibility. It is in theory the perfect ISA sound card. I just wish it had cleaner PCM. But it will do until the day comes when I can get an Orpheus 😁

The DOS drivers it comes with are a bit of a pain in the ass though. I believe UNISOUND covers this card so that's at least some relief.

I thought these chips didn't have SB16 compatibility (only some ALS cards) and that the OPL3 was bad? Interesting.

As far as I remember, the generic cmi driver is pretty straight forward. Initialization / and mixer tool with gives nice options like enable/disable 3D, Loudness, bass and treble adjustments if I remember correctly. FM is 100% Yamaha OPL3 clone to my ears, I made comparisons, both OPL2 and OPL3:

https://youtu.be/cZpzgJGZmnc
https://youtu.be/pjz9or5VqDQ

Cheers

Tiido wrote on 2021-01-08, 13:42:

I would simply get a suitable bracket or make my own. You only need one 374ohm resistor and one 90.9ohm resistor in addition to wires and connectors.

Maybe I am just being dumb but I cant find the referred image.. An image of the bracket or schematic would be nice so I can build one myself..

PARUS wrote on 2021-01-07, 21:09:

It does not matter. Both TTL and COAX are full-fledged proper S/PDIF. They differ in signal amplitude only. If you have TTL (2-3 […]
Show full quote

It does not matter. Both TTL and COAX are full-fledged proper S/PDIF. They differ in signal amplitude only. If you have TTL (2-3+ V, often 4-5V) and need COAX you can just use this:

That image is right here...

Anyway :

Marmes wrote on 2021-01-04, 21:13:

There are differences. I have both version A and C. Version A does not have 3D Sound on chip. While version C already has. I nev […]
Show full quote

There are differences.
I have both version A and C.
Version A does not have 3D Sound on chip. While version C already has.
I never seen a version B.
I guess HT1869V is just a relabel of version C.
These chips have good compatibility, but current design makes them noisy.
I guess there is no ADPCM Support in DOS , but that 's not really a big deal, most of the games don't use it, only a handful of them and those have patches or fixes to run without ADPCM, like DN2.
Overall is a nice card.

I have a "AV310" branded card which has the CMI8330A with supposed "3D On board"
I note it's not "on chip" as per your description but this seems like an odd choice.. (possibly the C rev was to save money??)

This card was an unexpected find in a Pentium 2 desktop I recently acquired and the card works great in a 386SX with DOS.
I have no idea how the 3D compares to the "C" revision chip but if anyone's got one and wants me to load a game and capture the output I'd be happy to.

I have recently tested a SoundPro 1869V+ integrated on a PC Chips M571 motherboard. It does appear to have 3D sound capabilities and its HRTF functionality works correctly in Thief Gold. Self noise is almost non-existent, possibly because it's an integrated solution. FM synth sounds identical to real OPL3.

You can read my full review in this thread. And I have uploaded a few FM synth music recordings here.

In my opinion situation looks like this:

CMI8330A - No 3D Sound
CMI8330A/3D - Has 3D Sound
CMI8330C - Has 3D Sound

@4xtx wrote about card with "8330A" chip and "3D Sound on Board" label. My guess that they (at C-Media) didn't remark a chip to "8330A/3D" but it is maybe a revised inside.

And SoundPro (OEM) lineup is:

HT1869V+ - No 3D Sound
HT1869V+ HRTF 3D Audio - Has 3D Sound (I don't know if they use 8330A/3D or 8330C chip)

Tiido wrote on 2021-01-09, 12:52:

PARUS wrote on 2021-01-07, 21:09:

It does not matter. Both TTL and COAX are full-fledged proper S/PDIF. They differ in signal amplitude only. If you have TTL (2-3 […]
Show full quote

It does not matter. Both TTL and COAX are full-fledged proper S/PDIF. They differ in signal amplitude only. If you have TTL (2-3+ V, often 4-5V) and need COAX you can just use this:

That image is right here...

Anyway :DirtyTTLcoax.png

It should have a pulse transformer to take care of grounding issues, or you can use toslink to give you optical isolation. SPDIF spec has little to no DC bias. Loaded voltage is fine at ~0.6V pp, but open circuit voltage will also be out of spec. A simple voltage follower as a buffer takes care of that, although make sure the op-amp meets spec for rise time.

Hence the "dirty" in the filename 🤣. Ideally there's the transformer as I mentioned several posts before, then it will be proper and without ground loop issues etc.
I have used these parts successfully for this task :
https://no.mouser.com/ProductDetail/710-750082154
https://no.mouser.com/ProductDetail/810-ATB3225-75034CT
https://no.mouser.com/ProductDetail/673-TM8116NL

Tiido wrote on 2024-02-23, 04:41:

Hence the "dirty" in the filename lol. Ideally there's the transformer as I mentioned several posts before, then it will be prop […]
Show full quote

Hence the "dirty" in the filename 🤣. Ideally there's the transformer as I mentioned several posts before, then it will be proper and without ground loop issues etc.
I have used these parts successfully for this task :
https://no.mouser.com/ProductDetail/710-750082154
https://no.mouser.com/ProductDetail/810-ATB3225-75034CT
https://no.mouser.com/ProductDetail/673-TM8116NL

Max typical clock rate for SPDIF I believe is going to be 192khz * 32 bits / subframe * 2 subframes for stereo = 12288000 b/s. IEC958 specifies higher sampling rates, but I've never heard of them being used over copper coax or TOSLINK. Rise time needs to be 0.4 IU, so the magnetics and any semiconductors have to handle 2.5 x the bit rate as a clock speed. That's 30 mhz and change.

Of course, if you're sticking to 44.1khz or 48khz as max sampling rate you can scale down your requirements.