Hi,
I recently had a "similar" problem to this.
I have a Bridgeport mill with the knee driven for the Z. One day the Z axis began to stall, exactly in the same point in the program, (so i thought).
The machine worked fine in manual jog, I spent ages zipping the Z axis up and down and could not get it to fault.
Increased the accell and velocity to bring it near the edge and still would not stall, but run the prog and I got tha nasty stalling sound.
I was sure it was the code, so I changed the code, slowing the axis at the fault line in the code, no difference.
The fault was so consistent "it had to be Mach"
Boils down to a sticky z axis screw, a good clean and the addition of a grease nipple and its been faultless for weeks now.
So back to the problem Michael has, I would take as much weight off the z axis, use a pen or a pointer to view the z distances and give it a try.
I might be miles off but if this can be tested easily, it might be a way forward, even if it eliminates and unlikley possibility..
Cheers
Derek.