ORIGINAL: HighPlains
Iron core motors do suffer from an effect known as "cogging". Certain design changes to the motor can limit the effect, but these tend to lower the maximum torque. A motor at rest tends to settle at a point between magnetic poles. What cogging means to the servo designer is that it takes a certain amount of motor current for the motor to climb the hill, and thus for a purely analog servo amplifier, a bit of error is needed before enough currect is generated to offset both cogging and friction.
Coreless motors do not have the cogging effect, so less motor current is need to start movement.
However with digital amplifiers, motor type is less important.
I don't disagree with the "cogging" theory but, when looking at the amount of distance between where the motor will stop between "poles" and then factoring in the gear ratio, the amount of servo arm position change would be so minuscule it wouldn't matter in most applications. If one needs that kind of accuracy then he needs to do like mrbigg says and spend the money to get something like a 5995 or equivalent. Also, let's remember the POT is attached to the output shaft and will request movement from the motor until it centers if the controller (amp) is set up right. In my humble opinion it all boils down to the quality of the whole package, motor, pot, board, etc. A good coreless motor with a junk pot or sloppy gear train/loose output shaft is still going to be a junk servo. Later Gents.