Mystery solved! Found an old post that mentioned the 280c logic board has 4 x 35v 100uf (check), 1 x 35v 47uf (check), 1 x 25v 33uf (check), and 1 x 25v 100uf (eureka!). I had all of the above in my parts bins, so recap complete.
My new friend gave me a bit of a scare today after buttoning it up though: Hung on the initial gray screen, even after numerous power-cycles. From my research trying to track down that cap value, I read numerous horror stories about leaking caps causing exactly that behavior on Duos
Gravely concerned at this point, I opened it back up and started poking around with the DMM and scope. I couldn't find a schematic anywhere and the board is a forest of vias (who knows how many layers...), but while probing around in the DC-DC circuitry, I noticed that "DZ6" was open! D = diode, Z = Zener I assumed. The little SOT-23 SMD had "8Y" silked across the top, and a bit of googling suggested an SOT-23 packaged Zener diode thusly marked was perhaps an MMBZ5248? That I certainly didn't have, but I contacted a tech I'm friendly with at a local electronics repair shop and he did Booted right up after replacement. A $0.13 part almost ruined my day.
This particular failure mode was probably specific to my board, but if any of you are hanging on to failed Duo boards, replacing DZ6 might be worth a shot...
I do this with my Macs - it prevents it coming loose when you transit the Mac, the video board socket doesn't hold onto the CRT's neck too well on its own from my experience.
Have you checked continuity from the brightness knob on the analog board?
@CC_333 Thanks - I did spend a couple more hours removing and testing everything, tracing through the schematic as I went. Then I noticed the yoke connection seemed a bit wobbly as I went to reconnect the board hoping that reflowing solder might have had some effect since I found no bad parts. I noticed
Someone has hot-glued the socket onto the tube!
And behind the insulating label I find this - note the big crack in the solder above the printed 'M' - the pad above 'I' was loose from the fibreglass but still electrically connected. I repaired these and now I have all the brightness I could want. But the front brightness knob still does nothing. I think there may be some reason the socket is hot-glued to the tube and it isn't going to be easy to find out what. Maybe if I gently hot-air it I can get the socket off without cracking the tube?
As for the floppy drive, head tension seems OK. Heads will still come together with no air gap when removed from the drive. I don't see how the spring could be moved to an alternate position for more tension either? Also I replaced the caps and no change. Tried another cable - a Segger J-Link has the same 20-pin cable as a Macintosh floppy - and still no good.
Actually, I'm 99.8% sure that's for the speaker?
This is a very incredible project! If anyone ever does one for the SE/30 (probably much more ambitious, but technically not much harder than this, I'd suppose? At least better schematics exist for it?), I'd consider giving one a try, as I have several boards which, if I haven't sold them all, I could salvage custom chips off of.
If there were a way to do it non-destructively (there probably is), I suppose one could even "borrow" the custom chips from a working board as well for testing purposes, yes?