Originally Posted by
dmatsumoto
Hi ikhan42,
The problem with rack and pinion, or anything that ends up meshing gears is that there's backlash. In other words, when you move forward in one direction by X, and then move back in the opposite direction by X, you're not exactly where you started because there is "play" in the interfacing between gear teeth. You can get around with anti-backlash gears that are basically two "slices" of the same gear, coupled by a spring that takes out the play. Other ways to do linear motion w/o backlash problems include ballscrews with anti-backlash nuts, linear motors, or belts and pulleys.
I believe one of the cool things about Mike's design is that it is a novel way to use belts and pulleys to remove backlash from linear motion, but it also avoids some of the weird resonant frequencies that would otherwise occur when you're traveling long distances with a typical belt and pulley setup. This makes motor tuning more complicated, and for me, tensioning multiple setups similarly with belts and pulleys is tough.
I haven't personally played with the servobelt, but if I ever come up with enough money to make a linear axis at home, this is the first thing I would go with, provided that it is okay to have the motor become part of the payload for that axis. At the moment, the only two disadvantages I see include that (i.e. the motor moves with the thing that it's moving, so it has to move its own weight as well), and the fact that all requisite wiring has to pass through the axis of travel.