This has been brought up at least a dozen times in this forums. Congrats! I think you are the first person who actually reports it having worked. However, maybe you are seeing a different type of banding that most other folks, since most everyone else has experienced banding from the start.
The type of banding described in the article is caused by a motor that doesn't microstep linearly, so an interference pattern appears when there is a mismatch between the thread pitch and layer height. As such, the height (pitch) of the bands will vary widely depending on your layer height, but go to zero/infinity asymptotically at multiples of 1/200th of the thread pitch. At a .3mm layer height, I believe I once calculated that microstepping-based banding should be about .5 millimeters high (about 50 bands per inch), or thereabouts.
However, the common type of z-banding that is inherent to the Solidoodle design has slightly different symptoms. It's always exactly equal to the thread pitch of 18 bands per inch. It gets progressively worse in amplitude as you decrease layer height (and conversely better as you increase layer height), but the frequency stays the same because it's tied to nonlinearities in the rod mechanism itself and not the motor.
Since you say this is a problem that developed after some use, I suspect this means either your motor is subtlely starting to fail, or maybe you just need to check and adjust the voltage that your controller board is sending to the Z-motor (see the wiki for instructions).
For anyone experiencing regular banding that is not 18tpi, I suggest using a voltimeter to check the z-motor voltages first before limiting your heights. After all, it's only by convention (encouraged by software) that we print at metric layer heights. Even with a metric Z-rod, you'd see the same type of banding at non-metric layer heights if your motor isn't microstepping properly. IMHO, it's better to keep full flexibility if possible.
