The following is a translation of the text from the source:
https://www.sblive.narod.ru/BAD_MEM_RETURNS/B … MEM_RETURNS.htm

For unknown reasons, the contents of the EEPROM (which contains device information) of Audigy sound cards gets corrupted. Usually the issue gets discovered when replacing the drivers or replacing the card. The main symptoms arising during this - the absence of AC3 Decode option in the mixer and the unwillingness of drivers to discover the card, when installing them from scratch.

If your card has been showing these symptoms, the only solution is restoring the contents of the EEPROM. But first we need to make sure the symptoms are caused by a failure of the EEPROM itself. For that we will need the program WPCREDIT - PCI Configuration Registers Editor by H.Oda (http://hp.vector.co.jp/authors/VA002374/src/download.html) .

NOTE: The link provided in the source seems to be dead. However, TechPowerUp seems to host a version of WPCREDIT:
https://www.techpowerup.com/download/wpcredit/

After launching the program select the option Device from the Edit menu, and you'll see a list of devices discovered on your system. Select the device "Audio Controller":

Next we select the option Structure View from the View menu:

The "Subsystem ID" line (marked green) may be different (see the table below). For SB0090 cards the number may be 0051 or 0053 (more on that later).

If the Subsystem ID line has nothing in common with the table above, either the contents of the EEPROM or the chip itself are damaged.

Usually cards with corrupted EEPROM show the number 0040 in Subsystem ID; the same number appears if you try to install a card with a desoldered EEPROM chip:

In my case, the Subsystem ID value was 00FF:

(to be continued)

(continuation)

The contents of the EEPROM looked like this:

Now, if you've discovered that the problem is the EEPROM, it is time to break out the old soldering iron. First we need to find the EEPROM chip on the PCB. Usually, among other letters and numbers, it is marked as 93C46. The chip itself is in the lower right corner of the PCB:

If you have difficulties desoldering chips like this, I suggest you cut off its legs with a scalpel and throw it away. The contacts are easily desoldered one by one. The following procedure can be performed with a new EEPROM chip.

Instead of 93C46, I recommend that you get the 93LC46 chip, which is pin compatible but has better characteristics. The type of chip package used on AUDIGY cards is called SOIC.

IMPORTANT - seems like AUDIGY cards do not work with chips that have 128 x 8 bit access organization (e.g. MICROCHIP 93LC46A), use chips with 64 x 16 bit access organization (e.g. MICROCHIP 93LC46B). Find the relevant information on the manufacturer website.

(to be continued)

(continuation)

You can write the EEPROM chip without a special chip programmer, you can use the LPT port instead and use the EEPROM programmer by Galim Mozhakov and one of the ROMs for Audigy.

The EEPROM programmer:

Here's how the EEPROM is connected to the LPT port:

I recommend that you launch the EEPROM programmer from pure DOS and set the LPT port mode to NORMAL (SPP).

The command line must look like this:

EEPROM.EXE [SWITCH] [ROM FILE]

E.g.:

EEPROM.EXE W SB0090.BIN

This will write the SB0090.BIN file to the connected EEPROM chip.

To verify the recorded rom you can run the following command:

EEPROM.EXE R TEST.BIN

The program will write the EEPROM contents to a file named TEST.BIN, which should be the same as the ROM file you recorded to the EEPROM chip.

Now we need to solder the chip back on the PCB.

(to be continued)

ROM files, part 1

SB0090.BIN original version, for SB0090 cards with metallic jacks.

SB0090R.BIN modified from SB0092.BIN, for SB0090 cards with metallic jacks. Not different in principle from the previous file.

SB0092.BIN original version, for SB0090 cards with plastic jacks.

(to be continued)

ROM files, part 2

SB0162.BIN for SB0162 cards without SB1394 interface (created by modifying SB0092 ROM).

(to be continued)

ROM files, part 3

DO NOT ATTEMPT TO WRITE THESE ROM FILES TO AUDIGY1 CARDS!!!

SB0240.BIN originel ROM file for Audigy2 SB0240 cards.

SB024X.BIN original ROM file (Subsystem ID - 1007) for Audigy2 OEM cards with plastic jacks, sent by Vadim Golovan.

(END OF TRANSLATION)

I'll add some of my own notes to this.

EEPROM corruption has been the bane of Creative cards and has been observed at least on Creative AWE, Live!, Audigy, Audigy2.

I would ask that people with working cards use the software provided here, the EEPROM programmer, to dump the ROMs for as many models as possible (EEPROM.EXE R DUMP.BIN).

I would also expect some tech priest out there to be able to build a less invasive LPT to 93C46 adapters, which would allow to program the EEPROM without desoldering it from the board.

I also wonder if socketing the EEPROMs would help restoring these cards easier.

What are your thoughts on this?

There's no avoiding desoldering the chip. These pins are actively driven and you cannot override them externally without causing discomfort (or outright harm) to everyone involved. I think the chipset is actually 3.3V based too and running 5V into any of the pins from LPT etc. guarantees the IO driver will die.

But since they're small SOP parts, it is very easy to desolder them, at least with the "knife" tip which can cover one row fully, making extraction just matter of seconds of effort. I don't think there are any sockets that fit the original footprint that you can hand solder however.

Tiido wrote on 2025-01-11, 10:37:

There's no avoiding desoldering the chip. These pins are actively driven and you cannot override them externally without causing discomfort (or outright harm) to everyone involved. I think the chipset is actually 3.3V based too and running 5V into any of the pins from LPT etc. guarantees the IO driver will die.

But since they're small SOP parts, it is very easy to desolder them, at least with the "knife" tip which can cover one row fully, making extraction just matter of seconds of effort. I don't think there are any sockets that fit the original footprint that you can hand solder however.

Well, that's a bummer. I wonder how the heat gun + kapton tape method would work.

Still, any solution is better than nothing.

93C46 serial EEPROMs are 5v tolerant / safe to program with a parallel port (assuming VCC is connected to 5V).
Ideally, if it's only intended to function as a ROM, you could replace it with an equivalent chip that has a dedicated WP pin to prevent against accidental erasure.

jmarsh wrote on 2025-01-11, 11:54:

93C46 serial EEPROMs are 5v tolerant / safe to program with a parallel port (assuming VCC is connected to 5V).
Ideally, if it's only intended to function as a ROM, you could replace it with an equivalent chip that has a dedicated WP pin to prevent against accidental erasure.

Yup, that's right. 2.5-5.5V range. Same goes for 93LC46:

As for an equivalent chip with dedicated WP pin, 24LC 16B seems to fit the bill:

https://ww1.microchip.com/downloads/aemDocume … t-20006054B.pdf

But will need to investigate further.

Pin 7 on 93C46/93LC46 is NC (not connected):

The same pin on 24LC16B is WP (write protect):

I don't know how that pin is handled on Audigy2, but connecting it to Vcc should inhibit all write operations. The Vcc range for 24LC16B is 2.5-5.5v. Connecting WP to ground should enable write operations (for EEPROM programming).

I think we may have found a permanent solution. Thanks, friend.

24LC16B is an I2C EEPROM, not SPI.
On closer examination it looks like 93C46 has a pretty non-standard footprint, the only likely substitute is a 59C11 (59C11 has a RDY/BSY# output pin) and they're functionally equivalent... both of them are commonly used in cars for security codes and odometer readings.

jmarsh wrote on 2025-01-11, 11:54:

93C46 serial EEPROMs are 5v tolerant / safe to program with a parallel port (assuming VCC is connected to 5V).
Ideally, if it's only intended to function as a ROM, you could replace it with an equivalent chip that has a dedicated WP pin to prevent against accidental erasure.

I didn't talk about the EEPROM but what it connects to, in context of "in-system programming" as suggested by a preceding post. Out of circuit you can do whatever with the EEPROM but still on the board, attached to a Creative chip on PCI, absolutely not (and still not on 5V ISA parts).

The EEPROM is a µwire standard thing, which is like SPI but precedes it. From what I know, SPI is modeled after µwire. In any case it is a standard EEPROM type and any programmer device can work with it, and numberous low-cost homebrew methods such as LPT connected things and even Arduino over USB type things.

It's an Soic8. At face value I don't see why a standard ch341a clip wouldn't work

(ie no need to desolder)

myne wrote on 2025-01-11, 17:36:

It's an Soic8. At face value I don't see why a standard ch341a clip wouldn't work

(ie no need to desolder)

Just because it's the same pin layout, doesn't mean the pin functions match (they don't). You would need a custom intermediate board between the programmer and the clip to remap them.

I can confirm that in circuit programming did not work for me.

If we're contemplating jamming things into printer ports, I don't see why 8 paperclips from the clip's plug wouldn't work.

insulated ideally.

I'll do some investigating and try things out on my own card, and document the process here. I also sent an email to Necroware, Scorp on Vogons, in case he wants to make a video about it.