Strange behavior with a continuous servo

AWOL:
The ex-servo almost certainly has analogue innards, which may drift over time/temperature.

Quite possible, does putting the writeMicroseconds inside the loop resolve this problem?