Avoiding Collimation Walk

During the process of collimating a Newtonian reflector's primary mirror, it is common to experience a problem known as "collimation walk". This is when the primary mirror cell creeps forward or backward over time due to a propensity to loosen or tighten the collimation knobs during collimation. 

Tightening the collimation knobs can draw the mirror too far back in the tube, which can bring the focal plane too close to the focuser, resulting in the inability to achieve focus with some equipment.
Even worse, over-loosening the collimation knobs can result in the knobs becoming disengaged from the mirror cell, and falling out. 

When this happens, it can be difficult to get the collimation knob back into place. Springs are usually used to force the cell away from the frame, and sometimes the loss of compression on the spring can push the cell too far away for the bolt to re-engage without physically pushing the cell back towards the frame. 

To avoid these potential problems, you can use just two of the three collimation knobs, treating the third like a pivot point and never touching it. This essentially locks the mirror cell in place so it cannot walk forward or backwards, since the act of over-loosening or over-tightening the two dedicated collimation knobs will inherently throw the collimation out of whack. While this can make the process of collimation a bit more challenging since some motions will require the use of two knobs instead of one, it will guarantee you won't have to worry about collimation walk. 

In fact, I designed my Dobsonian mirror cell and frame around the idea of two-point collimation to avoid the problem of collimation walk entirely: 

Mirror cell rides on the frame using a spherical bearing as a pivot point (top), and uses only two points of collimation (bottom).