Agree with not disturbing the optical alignment unnecessarily, but just in case (different make/model, but probably not too different): http://www.shtengel.com/gleb/Nikon_8000_9000_CDD_alignment.htm
Yes, I know that in some cases optical alignment is not too bad, but I also had experiences where a highly collimated device didn't work properly afterwards. This is a film scanner, and scans with 5400 dpi and the CCD has 3 degrees of freedom, plus there are a mirror and focusing lense, each with its own screws. And from what I gather from the scarce documentation, there are parameters to insert in firmware (with tools that do not exist anymore) for the final calibration after the optical path is aligned... Like you say, that would be last resort
They're not as special as you think --- and companies like Toshiba and Sony make tons of them, for both regular flatbed scanners (using a reduction lens) and film scanners (without the reduction); here's an example of a 3x7450 pixel, roughly 5400dpi sensor (using 5V and 10V supplies):
https://www.eureca.de/datasheets/01.xx.xxxx/01.04.xxxx/01.04.0122/TCD2716DG_090629_E.PDF
They're also not particularly expensive, as a search online for the above part number will show.
Yes, you are right, thanks. After I wrote that i did some more looking online, and I did find quite a few from Sony and Toshiba, but none yet with the right characteristics for mine. This has 24 pins (which is very common), in a DIP-like arrangement of 12 per side, but with a gap in the middle (i.e. 6 pins, big gap, 6 more pins on each side). I did find some with that pin layout, but none with the same arrangement (I'm checking the GND and 12V pins to see if they at least are in the right places). Still, if there's a CCD problem, the scanner will be resold for parts, since it would go beyond my ability to fix it (once again, optical path, but with a new sensor, there's no hope without the tools). Still, I'm satisfied that having 12V makes sense, since most of them seem to either be 12V, or 10V to 15V
One thing to be aware of if you're going to try testing with an external current-limited supply is power sequencing requirements; the CCD I linked above doesn't appear to have any, but if it's one with more complex active circuitry, you may need to be careful to prevent latchup (that is, assuming the regulator failure didn't already cause/was caused by other damage.)
That said, more pictures of the assemblies would be helpful --- maybe someone can recognise certain components and give you more info.
Agreed on the power sequencing risk. Assuming that the dead 12V supply has not caused problems, I would not want to apply 12V out of sequence, so I'm waiting for the 7812 to get here and use a current limiting resistor just in case
My main concern at this point is that I measure 88 Ohms between ground and 12V, with the 7812 disconnected. The CCD board is populated on both sides, and I can't see all components, nor follow the traces. My hope is that the 47 uF electrolytic capacitor is defective and partially shorted (yes, very uncommon failure mode, not that likey), or that 88 Ohms is for some reason expected... much more likely, though, I will replace the 7812 and discover that still it doesn't work, due to the failure of some other component on the CCD PCB. I'm enclosing a picture, in case someone has more ideas on what to check (I disconnected the main board from the CCD PCB, and I verified that the 88 OHm between 12V and GND is on the CCD PCB)
The current plan is to see what happens with a working 7812 and a current limiter and if it still doesn't work, I'll remove the main board that is blocking access to the CCD PCB and probe more. The reason I'm not removing the main board yet is that there are 2 flexible PCB cables and there's a non-negligible risk of cracking those (they are 15 years old), so I'm willing to risk connecting 12V to 88 Ohm for a short time and see what happens. I know it's not very smart