I have the 602. When you refer to the adjusting collar, are you referring to the clamp for the compound or the the compound lead screw drive components. If the latter, backlash is affected by three components. The lead screw nut, P/N 206; the thrust bearing assembly consisting of the lead screw mount, P/N 224, handwheel subassembly consisting of the handwheel, the threaded collar, and the set screw, P/N's 226, 230, & 231 respectively; and finally tightness of the gib.
The latter by virtue of a phenomenon known as "lost motion" Lost motion consists of backlash or play in mating parts and of physical distortion of parts due to applied force. As you tighten the gib, frictional forces come to play which require additional force to overcome. The parts start to buckle slightly, bearing preload increases, and the lead screw will begin to twist. You will see this as backlash. You can see the effect by loosening the gib so the compound slide freely. Measure the backlash with an indicator and tighten the gib, measuring backlash after each tightening. If you begin to see an increase in observed backlash it is due to this distortion.
There is a balancing situation for backlash. If the gib is too loose, lost motion will be minimized but the compound will move from side to side and will tilt with applied force. This is undesirable. On the other hand, if the gib is over-tightened, play will be minimal but the lost motion will increase as will wear on the moving parts. Tormach recommends tightening gibs to the point where lost motion just begins to increase for their mills.
Preload on the bearing is another area of adjustment. It is controlled by adjusting the threaded collar and fixed with the set screw
As to your question about alternate loosening and tightening when rotating parts, that usually occurs when the face of a part is not perpendicular to its axis of rotation. It can make adjustment more difficult but it shouldn't prevent you from making an adjustment.
