I just checked the datasheets and they seem to be pin compatible and electrically the same, but whether it's a simple direct transplant - I do not know..

http://dgmag.in/N27/crystal/CS4235.pdf
https://ftp.icm.edu.pl/packages/linux-alsa/ma … irrus/4237b.pdf

I just saw they mention a few differences with cx423xB chips but not sure if they apply... Check attachment. Hope someone knows...

If there's an eeprom on the board also (24LC04 or 24LC16), it might contain some configuration that's not optimal to the CS4237B afterwards. That can then be easily changed with the eeprom tools (resource.exe and the pnp.ini) available on vogonsdrivers. So far, I have only checked and swapped the CS4236 and CS4237b -which definitely are compatible.

MJay99 wrote on 2022-04-14, 09:55:

If there's an eeprom on the board also (24LC04 or 24LC16), it might contain some configuration that's not optimal to the CS4237B afterwards. That can then be easily changed with the eeprom tools (resource.exe and the pnp.ini) available on vogonsdrivers. So far, I have only checked and swapped the CS4236 and CS4237b -which definitely are compatible.

Haven't found any eprom in that area. I think the motherboard only has the bios and an eprom on the NIC. Now i only need to find a good source for those chips, any recommendations would be great.
Anyways if it works it would be really strange that such a bugged chip can be easily replaced by one that works as nice as the cs4237

marcushg85 wrote on 2022-04-14, 09:44:

I just saw they mention a few differences with cx423xB chips but not sure if they apply... Check attachment. Hope someone knows...

Did you just take a photo of your screen with a camera and post it in a forum?

kerop seems to know about the cs4237b:
Orpheus Soundcard: a new DOS soundcard with SPDIF/OPL3/MPU support

Cuttoon wrote on 2022-04-14, 11:54:

Did you just take a photo of your screen with a camera and post it in a forum? […]
Show full quote

marcushg85 wrote on 2022-04-14, 09:44:

I just saw they mention a few differences with cx423xB chips but not sure if they apply... Check attachment. Hope someone knows...

Did you just take a photo of your screen with a camera and post it in a forum?

kerop seems to know about the cs4237b:
Orpheus Soundcard: a new DOS soundcard with SPDIF/OPL3/MPU support

That's how I got to know about the cs4237b chip.
Is there any problem about adding pictures like that? Sorry if I did something wrong. I'll delete it if needed

marcushg85 wrote on 2022-04-14, 10:53:

i only need to find a good source for those chips, any recommendations would be great.

There might be cheaper options, but they're available on utsource for about 13 USD plus shipping and I also bought some from a seller in the US via the bay (322174185567) for about 15 USD plus shipping.
All of them were totally fine.

marcushg85 wrote on 2022-04-14, 11:59:

Is there any problem about adding pictures like that? Sorry if I did something wrong. I'll delete it if needed

Oh, to be young again! 😜
https://en.wikipedia.org/wiki/Screenshot

I don't think you'd destroy anything by attempting a replacement, but you may not get the best possible audio quality. In particular, I think the use of the filter capacitors is different between the parts. If you're set on trying this, it may be worth looking into one of the other parts, like the 4236 or 4238/9; one of those may be a better match.

Also, as MJay99 mentioned, look closely for an EEPROM containing configuration data for the part. They are usually just small 8-pin parts, and your picture does not show what U22 and U23 are. They may be completely separate from the sound hardware, but it would be better to make sure. If there is one, the data are mostly not interchangeable (with some exceptions), and may prevent a replacement chip from working (you'd be better off updating first). If there really is no EEPROM, it may be even worse; this could mean that there is BIOS/firmware-level code to do basic initialisation of the chip, and there is no way to know how it would react to finding a different chip.

Checking for the eeprom could maybe also be achieved by firing up the resource.exe from this package, after initializing the card with e.g. unisound successfully:
http://www.vogonsdrivers.com/getfile.php?file … menustate=46,38

After selecting the port, you could try and read out the eeprom and even let it print or disassemble (and even save) the contents if it's there. Just make sure to not write any default config to it, as that might really be difficult to recover from then 😉

I've checked the motherboard for possible small chips like those eeproms but couldn't find any (the ones near the cs4235 chip are amps).
Unisound detects a 4236 chip which is incorrect but I've tried to initialise the chip also with the original cwdinit and resource.exe can't find the eeprom I've attached some pics

MJay99

Did you also try Alt+A in resource.exe? That's the one that's always been finding it for me, even when everything else failed.

If still nothing is found then, I guess there really is none.

I'm a bit surprised about the CSC4236 ID unisound is showing for the card - I'd expect the CS4235 to report itself correctly and only a 'wrong' setting inside the eeprom maybe overriding it to CS4236.
But, who knows, maybe it's also a unisound issue (or it's due to what 640K!enough has been suggesting, that it's part of another BIOS/firmware init).

Sadly I also don't have the CS4235 here to try it myself. I was close to buying it, but in the end decided against it, due to the reportedly broken FM implementation, which I am now regretting a little 😀.

MJay99 wrote on 2022-04-14, 22:06:

Did you also try Alt+A in resource.exe? That's the one that's always been finding it for me, even when everything else failed. […]
Show full quote

Did you also try Alt+A in resource.exe? That's the one that's always been finding it for me, even when everything else failed.

If still nothing is found then, I guess there really is none.

I'm a bit surprised about the CSC4236 ID unisound is showing for the card - I'd expect the CS4235 to report itself correctly and only a 'wrong' setting inside the eeprom maybe overriding it to CS4236.
But, who knows, maybe it's also a unisound issue (or it's due to what 640K!enough has been suggesting, that it's part of another BIOS/firmware init).

Sadly I also don't have the CS4235 here to try it myself. I was close to buying it, but in the end decided against it, due to the reportedly broken FM implementation, which I am now regretting a little 😀.

Yes. I tried with alt A. Windows showed on the resource manager the Control part on address 120 but don't know if it has anything to do. Also tried but it doesn't detect anything.
I think i will try to buy just the cs4236... Maybe there's higher chances to get it working

This is a little strange, but it is nothing disastrous yet. First, let's address the basic questions:

1. Are you able to get sound in Windows?
2. Can you do your testing from basic DOS (no Windows loaded)? What does UNISOUND /CL report from straight DOS?
3. Before running anything else, what do you see if you try UNISOUND /c1 /r (again, from pure DOS is better)?
4. Do you have a valid CWDAUDIO.INI in the same directory as CWDINIT.EXE? Are you willing to post it?
5. From DOS, what do you see if you try CWDINIT /V /O?

This may give us some hints about what is going on. Some of these chips will identify as CSC4236 when no EEPROM is present, or the EEPROM is invalid, but not so bad that the chip fails to work altogether; it is a fallback feature of sorts, and relies on some level of software compatibility between the models to allow basic functionality to work.

640K!enough wrote on 2022-04-15, 05:57:

This is a little strange, but it is nothing disastrous yet. First, let's address the basic questions: […]
Show full quote

This is a little strange, but it is nothing disastrous yet. First, let's address the basic questions:

1. Are you able to get sound in Windows?
2. Can you do your testing from basic DOS (no Windows loaded)? What does UNISOUND /CL report from straight DOS?
3. Before running anything else, what do you see if you try UNISOUND /c1 /r (again, from pure DOS is better)?
4. Do you have a valid CWDAUDIO.INI in the same directory as CWDINIT.EXE? Are you willing to post it?
5. From DOS, what do you see if you try CWDINIT /V /O?

This may give us some hints about what is going on. Some of these chips will identify as CSC4236 when no EEPROM is present, or the EEPROM is invalid, but not so bad that the chip fails to work altogether; it is a fallback feature of sorts, and relies on some level of software compatibility between the models to allow basic functionality to work.

I checked everything you mentioned in pure dos.
Sound is perfectly working (with bugged Fm) under pure dos and windows.
Interesting that CTR:FF8 but don't know if that is the adress for control
Anything else?
Thanks for your support

My suspicion, then, is that no EEPROM was included. This is presented as an acceptable design in the documentation for the CS4235, even if it isn't something that I would want to do. I guess it shaved a few dollars total off of the BOM, so they shipped it that way in spite of the drawbacks.

Given that it still identifies itself as a CS4236, it doesn't appear that there is any fancy BIOS/firmware-level initialisation taking place, beyond basic Plug and Play resource assignment. Based on your photos, it appears that the control port is indeed FF8H by default (before you run UNISOUND or CWDINIT). If you are not using the /O option when using CWDINIT, the control port probably stays at FF8H, in spite of what is specified in your CWDAUDIO.INI.

So, if you attempt the swap, you may need some additional help to initialise the new chip to get the best results out of it. The important thing, if you don't want to start re-arranging support components on the board, is to ensure that you find a version of the chip that most closely matches the design specifications of the CS4235. I don't think you will be able to get the best that the CS4237 can offer on that board, because of the differences in the way the two handle the reference voltages and filter capacitors, and maybe a few other details.

Thans for pointing to Vogons driver package including EEPROM image for 4236B. I used this chip instead of 4237B for Rasteri's PC/104 soundcard because I have a spare one. I just assembled the PCB and trying to bring it up so I hope it will work after proper EE upload...
Interesting that is also includes a FW patch - probably it's dumped from some existing soundcard. Are there known more FW patches versions and haw it can be recognized what's newer?

interesting, thread resurrected after so long, I just got some of those 4237b chips i ordered long ago and will try the transplant to see if it works, will report when i do it

So, Finally after one year, I did the mod and replaced the cs4235 chip with the cs4237 and... it didn't work...
Chip is detected but not working on dos and in windows it gets detected as Crystal Codec but can't install any drivers, I guess the lack of an eprom makes it not to be able to give any resources,
I tried many drivers and unisound but nothing made it work. So eventually I tried to add into the computers bios the whole info of cs4237.ASM (firmware) into some blank space in the bios content but of course, that just bricked the computer... and now I don't know if ill be able to recover it... as the computer doesn't have a normal bios but that strange dual firmware to be used as thinclient or workspace on demand (regular pc behaviour). I made a dump before the modification with uniflash but could be wrong.