Happened with two different servos? Might this be a problem with the clock crystal or related circuitry on the EBB board? The timing of the pulse width controls how far the servo arm moves. And the timing is controlled by either an external xtal or one internal to the uprocessor. If it is slewing for some reason, then might there be such behavior? I suppose if the pin on the PIC generating the PWM is impacted there could also be a problem (e.g., some other problem resulting in high frequency interrupts on the chip, robbing time from the interrupt responsible for managing the PWM duty cycle causing late issue)?