But with respect to #1 is the diag sketch still reporting that all the tests passed on
the serial port even though the displayed turned off?
The reset line on these ks0108 glcds varies between modules.
Some modules have pullups on them, some don't.
Many seem to self reset when powered up. But some don't and require
an actual pulse on them to function properly.
Ideally, it should be driven and it should be pulsed low to ensure the glcd
is put into a consistent sane state prior to initialization.
But that takes another AVR pin.
On many glcds you can get away leaving the reset pin unconnected,
but it is best to tie it high.
On some you have to actually pulse the reset.
I wrote about the different ways of hooking up the reset line
on the ks0108 playground page. So see the playground page for more information
about how to handle the reset line and potential issues and work arounds.
My guess is that module you have is experiencing a spurious reset
while the test is running because the reset signal input is not driven high.
Many ks0108s don't enable the pixels after a reset so what you are seeing
is probably a spurious reset because the reset input to glcd is floating.
I'm guessing that the diags would report a failure when this happens and that is why
I'm curious to know if you see a reported failure on the serial port when the screen pixels turn off.
The reason the pixels turn back on with the diags is because each test iteration re-initializes the glcd
and part of the initialization (it is actually the first thing done) is to turn on the pixels.
--- bill