There are only a couple mentions of this on the net, one saying "it's a capacitor problem" (but, frustratingly, not saying _what_ capacitor), another (regarding the STR-DA30ES) saying it was a bad solder joint on the bridge rectifier feeding the +5V line.
The service manual has complete circuit diagrams and component board traces, but nothing about the firmware or error messages. I worked out the fan protection circuit, and worked out that I could tie the thermistor monitoring line to +5V to trick the controller into thinking the temperature was okay (working on the theory that a capacitor had shorted out). This let the relays click on, but the display would still flash PROTECTOR (FAN).
After finding the STR-DA30ES mention, I probed the bridge (D955). I got about +1V on the DC side -- it should have been around +9.7V. This line feeds a couple regulators to make the +3.3V and +5V lines, but they too were less than 1V. Weirdly, the AC side of the bridge had 0V. The AC line from the transformer had around 9.7V. Narrowing in on the section in-between I found that fuse F953 on the main board had blown. I'm not sure why there was +1V on the DC side of the bridge (though there are multiple power circuits in the system), but the low +3.3V and +5V lines are apparently enough to piss off the controller chip, and I guess the thing it first notices is the wrong voltage on the thermistor monitoring wire.
Shorting over the fuse made it work. I'll hit radio shack this weekend for a new fuse.
So, dear future reader who found this through a net search, check fuse F953. It's on the main board (the big one along the bottom), in the front corner that doesn't have the fan connector. Glass cylindrical fuse.
(And to my puzzled friends who watch my livejournal: hey, this was an easy way to publish something for the search engines to find. And I guess the moral of the story is, check the obvious things like fuses first.)