The transistor like part should be a 78L05 (is it ?) which is a voltage regulator, and if there's 12V on both sides the main OPTi chip is dead and possibly other things too... It is supposed to make +5V for analog parts of the card and 12V will certainly kill the main chip though opamps etc. are likely to survive as most take up to 30V before dying.

its a LM78L and pin 3 is tied directly to the 12 V, so i think its ok. The chip appears to just regulate to 100 ma.
I was just measuring voltage and the 5 V looks ok. I checked on the pdip chip which is 93LC46 eeprom chip.
I was checking the -12 V line and that isnt giving a good reading. So ive been trying to trace that around.

Edit:
I think the -12 V might be ok, i get -12 near the edge connector. It does end up at the HEF4052BT and i mearured right was around -10 V.
I also found the Vcc and Gnd test points and that shows 5V.

If 5V comes out one of its pins then that part is fine and plot thickens.

I'm actually not completely sure what is the problem...

When card is alone in the system, is it being detected properly and set up or always fails ? Chances are the EEPROM got corrupt (happens occasionally on AWE64 cards). If it is set up, using Duke3D setup, will it detect MPU401, Adlib and will it be able to do SFX ? That setup can be quite useful in determining problems.

so short answer is that it was the eeprom. The card is working again and unisound detects it. Now it took me a bit of a process to get there.

Today i unsoldered the eeprom chip and checked it in the reader i have and it had bogus data. BB's and FF's. But i could write anything to the chip so I knew it worked.
Along with duke3d Id thought id try to opti931 driver package also. The driver package is the one from vogons. I just ran the setup within win95 and that gave me the dos utilities.
I found i could run the sndinit /b or sometime just sndinit and the card would configure and everything worked, but not across power cycles (eeprom in question at this point)
There was another program called sndtune.exe and this was the key one. It did something and then rebooted the machine. It came back up and unisound saw the card again. Dumped the eeprom and it had real data.

So the question i have in my mind now is if unisound caused a corruption or if i was something else i did. One suspicion i recall is there was a jumper and at one point i did try to use it (it disables the PNP function based on the spec). But it didnt work (i would need the sndinit) so i removed it. It was working after that point.
I will do some more testing and see if anything i did with unisound causes again, but i really glad i found a way out of the hole.