Solved it, it was the adaptor. That's what one unsoldered pin does... Thanks anyway!

First of all, thanks for all of your hard work on this!! I discovered this thread probably about a year ago when I noticed the hanging note bug with my AWE32 CT3900 while trying to record some game music. I finally worked up the courage (and gathered the necessary equipment) to desolder the DSP chip and replace it with a socket yesterday, and so far so good! I have some questions, though:

1. What is currently the best firmware version to use for the CT3900? (Assuming that I have no need to play any Duke games with it.) Do I understand correctly that v4.17 is based on v4.16, which normally doesn't work with the AWE32, but v4.17 has changes that make it compatible? So is v4.17 the best right now? Or v4.13 patch 5? I tried both of these, and they both seem to work, but I didn't do a lot of testing and didn't notice any difference.
2. When I tried to write the latest v4.17 to a spare chip, I ran into an error that the file size was too big (8256 bytes when it should be 8192). Granted, I'm on Linux and using an open source program (https://gitlab.com/DavidGriffith/minipro) to interface with the T48 programmer rather than the official Windows program, so your mileage may vary. Seems like a mistake, though. I looked at the file in a hex editor and saw that the end of it was filled with junk 0xff bytes, so I just truncated the file. It should probably be fixed in the repo, though.
3. Looks like DMA clicking is still a known issue? I think that's what's happening in Star Trek Judgement Rites. The recorded audio in cutscenes and voiceovers is pretty glitchy. I'm wondering if it could be related to what you're seeing in the Duke games?
4. Sorry if this is only tangentially related to this thread, but I have an AT89C52 in a DIP40 package that's read-locked that I need to dump the firmware from. It's similar to the DSP chip which you all have been able to get dumps from, so I'm wondering if someone with the necessary skills and equipment could help me out? I'd be willing to pay for their trouble. I'm fine with destroying the chip in the process; I specifically got an extra for this purpose. It is in the spirit of preserving old hardware; it's from an Avant Stellar keyboard, and it would allow repair and modification. I've reached out to some people who worked at Creative Vision Technologies and the company that bought them to see if anyone had a copy of the firmware that they'd be willing to release, but I was told that it does not exist anymore. So getting a dump from the chip is my only hope...

Keep in mind I have an unresolved issue with using an aftermarket DSP chip on an AWE32. Sadly my particular awe32 did not have sufficient room around the PLCC footprint to fit a socket, so I wasn't able to troubleshoot deeply, but I was getting spurious NMI generated from the Atmel '52 chip. I find it unlikely to be the code and more likely to be the chip itself, but I was never able to definitively determine the cause. I just reverted to stock DSP.

4.13 patch 5 gets my vote

maxtherabbit wrote on 2025-10-08, 23:15:

Keep in mind I have an unresolved issue with using an aftermarket DSP chip on an AWE32. Sadly my particular awe32 did not have sufficient room around the PLCC footprint to fit a socket, so I wasn't able to troubleshoot deeply, but I was getting spurious NMI generated from the Atmel '52 chip. I find it unlikely to be the code and more likely to be the chip itself, but I was never able to definitively determine the cause. I just reverted to stock DSP.

Which AWE32 do you have? My CT3900 had some nearby components as well; I ended up just putting the socket pretty much on top of them... Hopefully it won't cause any issues. I'm using an Atmel AT89S52, and so far no issues that I can tell, but is there a way for me to check for NMIs?

I attached some photos I took of the process; are the component locations similar to your board?

Mine is a CT2760 rev 1. The components on your card are slightly further from the footprint than mine are. As for checking for NMI, you would get a "PARITY ERROR SYSTEM HALTED" screen if you have parity checking enabled in BIOS (or maybe even if you don't)

Am I correct in thinking that there is no fix for any Awe32/SB32/Vibra that doesn't have the CT1741?

I don't think so. I believe that the DSP is part of the same silicon as the audio chip. This project requires that the DSP be on a separate chip, since we are literally replacing that chunk of programming logic in ROM.

meyertime wrote on 2025-10-08, 22:25:

4. Sorry if this is only tangentially related to this thread, but I have an AT89C52 in a DIP40 package that's read-locked that I need to dump the firmware from. It's similar to the DSP chip which you all have been able to get dumps from, so I'm wondering if someone with the necessary skills and equipment could help me out? I'd be willing to pay for their trouble. I'm fine with destroying the chip in the process; I specifically got an extra for this purpose. It is in the spirit of preserving old hardware; it's from an Avant Stellar keyboard, and it would allow repair and modification. I've reached out to some people who worked at Creative Vision Technologies and the company that bought them to see if anyone had a copy of the firmware that they'd be willing to release, but I was told that it does not exist anymore. So getting a dump from the chip is my only hope...

Hi Meyertime. If you want to get that firmware dumped you might want to contact Tubetime / Eric Schlaepher. Apparently he's on Bluesky at https://bsky.app/profile/tubetime.bsky.social . He's the guy that reverse engineered the original Soundblaster 1.0, created the SnarkBlaster clone and dumped the Soundblaster 2.0 ROM.

So, uh, Single Cycle DMA clicking still going on?
I have a lot of different AWE/SB32 models so I'm very interested in this and maybe I could help if you need something tested or a certain FW dumped.

Would be too cool if that CT3980 finally became officially bug-free, it's my favorite sound card of all time.

Also would it be possible to just transplant an unaffected DSP from, say, an SB32 non-AWE (CT3670) or an AWE 64 Value (CT4500)?
Both models are unaffected by the Hanging Note Bug and the Single Cycle DMA clicking.

Danger Manfred wrote on 2025-11-03, 22:34:

Also would it be possible to just transplant an unaffected DSP from, say, an SB32 non-AWE (CT3670) or an AWE 64 Value (CT4500)?
Both models are unaffected by the Hanging Note Bug and the Single Cycle DMA clicking.

On awe64 the dsp is all integrated into one giant chip. I believe the CT3670 is also an awe64 in disguise. it has integrated dsp as well.

I bought some of thoose chips on AliExpress (following a link from the Bit's & Bolts Website) and was unable to flash them. Bought some more from other sellers on AliExpress and now ended up with three boxes from the thee different sellers virtually looking the same. And the chips don't look right to me, some are flat all over, some have one ore two round deepenings in the corners, some don't. The print also looks suspicious. Are all of this fakes? Or is maybe my adapter for my programmer (GQ 4x4) to blame?

Curios where others got there chips and the experience they made. 😉

I bought new production from digikey

Cheap electronic parts from AliExpress are not necessarily a fake. In most cases they are legit but roughly recycled from scrap pcbs and though may not just work due to a heat or electrical damage. Usually seller just doesn't bother to check their condition and hope that at least some of the minimum bunch of ICs sent will work.

janosch wrote on 2026-02-16, 20:27:

I bought some of thoose chips on AliExpress (following a link from the Bit's & Bolts Website) and was unable to flash them. Bought some more from other sellers on AliExpress and now ended up with three boxes from the thee different sellers virtually looking the same. And the chips don't look right to me, some are flat all over, some have one ore two round deepenings in the corners, some don't. The print also looks suspicious. Are all of this fakes? Or is maybe my adapter for my programmer (GQ 4x4) to blame?

Curios where others got there chips and the experience they made. 😉

Have you checked your adapter board to make sure it is working fine? It could be a bad socket etc?

I buy chips from AliExpress, no problem. Just be sure to test them ASAP and ask for a refund on the duds. Super easy. It's only problem if you buy in bulk and don't test them until after the return window has closed.

appiah4 wrote on 2026-02-17, 06:20:

Have you checked your adapter board to make sure it is working fine? It could be a bad socket etc?

Yeah, I checked every single pin in the socket for continuity to the pin, that's ok. I just don't know if the adapter is really the correct one, pin1 is in the middle of the top is that correct?

janosch wrote on 2026-02-16, 20:27:

I bought some of thoose chips on AliExpress (following a link from the Bit's & Bolts Website) and was unable to flash them. Bought some more from other sellers on AliExpress and now ended up with three boxes from the thee different sellers virtually looking the same. And the chips don't look right to me, some are flat all over, some have one ore two round deepenings in the corners, some don't. The print also looks suspicious. Are all of this fakes? Or is maybe my adapter for my programmer (GQ 4x4) to blame?

Curios where others got there chips and the experience they made. 😉

I got mine off eBay (from China). I found these were the best value since I have multiple cards that need new DSPs. If you're OK with buying 20 at a time (seller has lower quantities available as well, but the price per goes up), I had OK results with these.

https://ebay.us/m/OC6bqE

18 of 20 worked. They were different models mixed in between the 24JI and 24JC ("I" actually being better as its industrial quality) and they certainly were NOT new, but the price was good and I was happy overall. YMMV of course.

1 test adapter with known working micro
2 can those micros be write locked? some cplds can be locked after write requiring special parallel programmer to unlock

Are these chips write once or can they be written to and then erased (easier for testing)?

The Atmel AT89Sxx are rewritable many many times, but there can be OTP versions of some compatible chip. There also isn't a problem to erase and write it even if locks are set, because erase also removes the locks. But if the chip fails to get detected etc. it is probably just defective or maybe not even the chip at all. The ones on the photo seem to be a bit suspicious...

As far as adaptor goes, if it is the one meant to be used to write these MCUs, it shouldn't be a problem but I am unfamiliar with that programmer and its adaptors. Mine requires a completely different adaptor for PLCC44 action, and it connects up all the pins too.