I have never seen anyone completely gut a servo's electronics and drive the motor directly before, there's really no point, that's why servos are so nice, the drive electronics are already there, use them, I'm not sure what Nicksan was talking about; Modifying a servo for continuous rotation is as simple as detaching the feedback POT from it's drive train and using some hot glue to lock it in place. Then instead of the servo pulse relating to a proportional position it becomes a proportional velocity (torque) The further from center (whatever pulse position the feedback pot was locked at) the pulse gets the higher the torque pulses the motor gets.
What you've described is a servo modified for continuous rotation. The only bad thing about this is if the servo electronics are under extreme operating conditions (temperature or voltage) the normal servo 'center' signal will cause the motor to drift juuuuust every so slightly in one direction or another. Not an issue if you just don't send the servo a signal when you're not driving it, then the driven wheel will just float.