Fortunately, it turns out to have been much simpler.
The 820-II firmware
apparently expects the FDC controller to be present during init. To keep
things simple, I had the daughterboard unplugged.
Once I smartened up and plugged the controller in, it started responding
to the keyboard. I can only assume it resets the controller at power up
and is waiting for some sort of state change or interrupt. Without the
board present it was just waiting forever and never getting to the stage
where it would acknowledge character input from the keyboard.
This reminds me in a way of a problem I had with a Whitechapel MG1 (A
32016-based workstation). It flashed the error LED at power-on, reporting
a multi-bit DRAM error.
TO gcut a very long story short, after battling throuhg the memory
arbitration logic, I discovered there was actually nothing wrong. The
mainboard has 512K of RAM on it, the boot ROM needs (and tests) 1.5M.
After plugging in the 2 RAM expansion boards I'd removed 'to keep things
simple' (:-)), it worked fine. This is not documented in any of the
manuals I have.
Nice when things turn out this easy.
In my case it wasn't so simple. It took me a day to go through and check
all the logic. It's a complcated machine...
-tony