The ROM should be cold and not even warm, so it seems the problem is found. It likely has a dead IO driver and that makes one signal stuck and that ends up corrupting everything on the sample bus and explains why things don't work out.
I have a spare chip somewhere, if I can find it, it can be sent your way since it will just continue to collect dust 🤣. (I moved recently to Norway and much of my stuff is still packed).

oh - I thought we established that most of those chips can be hot! 😀
I did check the lines with my scope, I haven't finished but I did observe some weird stuff on some lines as if there was some bus conflict (signals at 2.5V and so on)

And that ROM is not a regular one that can be sourced and programmed by myself? In any case, if you end up with a spare that would be cool. I'd pay for postage of course.

One more question (Columbo mode ON)! 😀

When playing games like DOOM, the soundtrack is working fine. Doesn't that require that ROM/sample memory?

Thanks again and I hope you find the spare chip! 😀

Tony

On that particular creative card, only the speaker amp chip can be hot, and only with 4 ohm speakers and significant volume for a while 🤣. Original not-integrated EMU8000 chips get very hot, but the integrated stuff on this particular card won't. DRAMs are barely warm in normal use and the ROM should be cold pretty much.

Those 2.5V lines are absolutely not normal, that should never happen on the sample memory bus there.

DOS games through AWEUTIL or ones with native AWE32 support use the ROM bank, but if those sound right then things are strange... but if some games are good and some not, the plot thickens 🤣.

I am unsure when the chip will get uncovered, I don't have a workspace yet and not even space to unpack many things. I am not yet sure how much will shipping cost over here...
AM29F800 could be a possible replacement, though it has several flash specific pins that require special care that the ROM will not have.

DOS games through AWEUTIL or ones with native AWE32 support use the ROM bank, but if those sound right then things are strange... but if some games are good and some not, the plot thickens

Yes, that's why I was puzzled. DOOM works, MIDI works. It's the "synth" in Windows and the AWE Synth from Diagnose which sound distorted.

I'll try some more games to test.

No pressure about the replacement chip - if and when you can find it, please do let me know!

Back to this after 4 years, I made a video where I repaired an AZtech card and I used some parts from this SB - a viewer mentioned the SB could be recovered as the EMU8011 is a ROM chip. And I kind of remembered the EMU was getting very hot.

Came back here and indeed someone suggested the same but then time passed 🙂

I wouldn't mind attempting an EMU8011 replacement. AI is telling me this is a regular UM23C8100 or MC23C8100 chip in SOP-44 but I cannot find any. Is there any info on this chip? Is there a dump? Can someone confirm the chips I mentioned would work?

Thank you!

1 and 2MByte sized flash chips will mostly share the pinout (there's a difference around the write pin which is missing on ROMs), so it wouldn't be too difficult to adapt them for this task.

I might have a chip or two in my stash, which I can mail over if I find them and cost for that gets covered.

EDIT: I got access to my stash and there is one chip that you can have if you really want. I do think some mechanical problem is much more likely to cause issues however, such as a bad trace or a via or maybe even a loose pin.

Hi Tiido

While I do appreciate the offer (I honestly do, I am not being sarcastic), we had this exact same conversation in 2022, just a few posts above 🙂

I really appreciate your help but if you could point me to the right direction by mentioning a chip which would work, I'd be happy to source it here myself 🙂

🤣, yeah... blast from teh past I suppose.

I think there may be a direct compatible, but only if you manage to source a proper NOS chip since it is an OTP EPROM, which you still have to program yourself with the correct dump. If it isn't a true NOS part (meaning it is programmed) you have no way of reprogramming it.

The part is M27C800-xxM1 where xx is a value from 50 to 150. I'm not sure what speed rating is necessary for EMU8000 however. Flash chips will have the !WE pin that has to be taken care of so a board modification will be necessary...

I'd imagine NOS = New Old Stock. So "never programmed before" because it can be found in OTP (One Time Programmable) version, so if it's not new, it'll be as useful as a chocolate teapot. There is also an UV-erasable one though. I do have an UV lamp.

Ok so the part is an 8Mbit ROM.

What about the !WE pin? Is the process documented somewhere?

The UV erasable one is only in DIP, all other packages have no window and therefore cannot be erased. My quick seach also suggests the SOP44 version is unobtanium, it wasn't common to begin with... but you can try to locate MX29F800 flash or some other 29F800 part. Aliexpress seems to have this part available, then you only need a dump and write the chip.

The basic idea is that flash chips (29Fxxx of some sort) have a pin to control writing to the chip, that ROM and EPROM chips do not have since they are not writable in-system, and it is usually in place of a address line. You have to disconnect that address line and rewire it to where it is supposed to go, and connect the !WE pin of the chip to high state to put it into permanent read mode. Looking at the board photo above, there is an extra address line connected to the pin that is !WE on flash chips, that is actually not used on the ROM itself, perhaps for possible future 2MB ROM model that never materialized... So you have to cut that trace, and connect the pin of the chip to VCC (or leave the trace connected and cut it some location where it goes near VCC and then bridge it there if it can give a more desirable result).

Looking at the pinout of MX29F800, there's also an extra !RESET signal that has to also connect to VCC but it is right next to the !WE pin so a solder blob can connect those together. Stealthy mods are possible with some effort though.

That sounds doable - is there a diagram for the pinout of the EMU chip? So I know what to wire and where.

It has a normal JEDEC standard pinout, you see equivalent in datasheets of EPROMs etc. in same package.

Gotcha - then I am lost. The EMU chip and the MX29F800 have the same number of pins, right? But you say the MX has extra pins which will need to tight high.

You have to disconnect that address line and rewire it to where it is supposed to go

If both chips follow the same pinout, I don't understand this.

The amount of pins is same, but pin 1, and pins 43 and 44 have a function on the MX while on the original mask ROM they are No Connects (but pin 43 is used on the PCB for A19, to support 2MByte/16Mbit ROM chips, which never appeared, and you see it collides with the !WE signal). These 3 pins are address lines on larger capacity ROM chips.

Thanks

I've attached a tentative pinout.

So the plan for the MX chip is:
pin 44 - disconnect and pull high
pin 43 - disconnect and pull high
pin 1 - Ready/Busy state - does it matter?

On the PCB
pin 43 - pull to GND just to avoid issues? This is unused for larger chips which never happened
Pin 44 - leave it as is as NC
Pin 1 - what is the board expecting on pin 1?

Thanks

Pin1 on the board is unused, and the Ready/Busy pin on the chip is an output so nothing has to be done with it at all. That pin only reacts during programming and erasing.

Pin44 and 43 are to be linked together and connected to VCC, with trace on the board to pin 43 getting cut. 43 in an NC on the ROM chip, it is only 1MB in size and only a hypothetical 2MB chip would use it. That trace you leave alone after cutting since it is an output from the EUM8000 part of things and tying it somewhere will hurt the IO driver in the chip. You can do the cut at location 1 and/or 2, it doesn't matter very much.

thanks. Then the final mod is:

MX chip
pin 44 - disconnect and pull high
pin 43 - disconnect and pull high
pin 1 - Ready/Busy state - just ignore and solder on pad

On the PCB
pin 43 - leave as is, just disconnected from IC leg
Pin 44 - leave it as is, just disconnected from IC leg
Pin 1 - connected to leg 1 of IC

I'd rather not cut traces so I'm exploring the option where legs 43 and 44 are lifted, joined and pulled to ground.
On your drawing, I see you're connecting legs 43 and 44 to BYTE#, is that equivalent to VCC?

What about the IMAGE to flash on the replacement chip? I believe it's still copyrighted. Can it be found online?

I find cutting traces to be much preferred over bending pins, since the pins are a lot more fragile and when one breaks it takes a lot more effort to get things to usable state again...

!Byte signal connects to VCC to select 16bit mode over 8bit mode.

I don't know if the dump is publicly available but I have done one.