Originally Posted by
rzrostlik
Phil,
I believe that using a two-step homing process will improve the situation if the positional repeatability is due to the switch (and not backlash in the ball nut).
In the two-step process, we are actually using the index pulse of a shaft encoder to represent 'home', rather than the position of table is at when the limit switch changes state.
The assumption here is that the index pulse on an encoder is much more repeatable than a mechanical switch changing state. This process requires that the switch is repeatable within one full revolution of the shaft encoder. This is usually not a problem. So, we use the mechanical limit switch to trigger an event that reverses the stepper motor and start looking for the very next encoder index pulse. Once the index pulse is found, we call this 'home'. Everything is referenced from this location, not when the limit switch changes state.
The beauty of this approach is that you can have a fairly sloppy limit switch and it should not impact home reference repeatability.
Rick