Well after nearly a year of searching for lost steps, which of course took untold number of hours and frustration, I happened across this link http://www.candcnc.com/G251-4_FMK.html. 

When testing my z axis one thing I did was write a simple file, as I discovered today is suggested, that would have the axis change direction multiple times.  When ran the file I discovered that with each direction change there was an equal amount of position loss.  I can't remember now but about 20 cycles of direction change on z would equal about 2mm of position loss.

Thought I should post in case there are others out there.  This isssue is likely an old one but always best not to assume.

S

Good to hear, hopefully you will be sorted once you get it done.
Hood