AVR-Chat

>> The period between the pulses can be varied, shorter means more
>> accurate control.

I worked with this a while back on the 8515.   With a longer delay
between pulses, the servo wasn't active as much, and with extreme
values, it turned into a "bang-bang-bang" kind of output, with each
pulse bringing the servo closer to proper position, and not driven
most of the time.  With high rates, the servo was noticably stiffer.
I didn't measure it, but I suspect that the motor is only driven for a
short while after the pulse stops.

I also found fairly large variations in the pulse widths for
"standard" servos.  IIRC 750uS was the bottom end.  I had a bunch of
different brands, and ended up implementing a table so that I could
set the servo brand per channel, and calculate the widths based on
what I measured for min and max on each.

1ms should be in-range for most anything.  Maybe 1.25 would be safer.

Something I didn't mention before, and I should have, is that you
don't want to hit the servos with pulses that are out of their
operating range.  They will run into the mechanical stops, and this is
hard on the motor and gears.  You can do it for short times, but  it's
best not to do it at all.