repeatability and calibration issues

I Just got everything set up on a new build: 7i76e + linuxcnc + integrated servos that accept step commands with built in encoders(they do not loose steps).

Problem repeatability! If I zero the machine then issue g0 x60 ..(60") some times it goes to 60" and some times its up to .25" off then g0 x0 some times it comes back in the right spot and some times its up to .25" off
notes:
*the machine has rigid ball screws with 0 backlash. (rm2005 5turns=1")
*there is 0 mechanical backlash (have used this machine for years fine on g540+steppers never had a backlash problem and
there is still no play, checked the ball screws and axis for any mechanical play again and there is still none.
*I thought this might be a micro-stepping issue how ever the pin-conf wizard does not show a setting for this like the step-conf
*I did not know what kind of drivers the 7i76e uses so I selected custom

If you are certain that you don't have any mechanical issues (slipping coupler or pully).
Then it is lost steps from having too short step timing settings or electrical nose are the most likely suspects.
What are the minimum timing settings for your servo drives? (It is often wise to double (or even triple) the minimum step length and step space times given in drive's documentation. Other timings can often be set even higher without ill effects.

thanks for the response. The couplers do not slip, clear path motors are integrated drivers servos and encoder within the motors they will not miss a step if the step command is actually being sent. they claim to be 1us responsive i have them at 5.
could this be a maxjiter issue

In the ini file you posted you have specified 1000ns (1us) as the step timing. I would say that is the source of the problem. Sure the servos won't miss steps but only if they can see the step. Are you using differential or single ended signals?

Did you try the 5us timing after you posted your ini? Did it work any better? If you did try the longer step timing and it didn't work. That leaves a wiring/electrical problem is probably interfering with your step signals.

Another possibility is wrong step polarity. The 7I76E STEPN+ outputs are active high
and the STEPN- outputs are active low.

As a general rule I would start with step time and step space at 5 usec (5000 ns)
and say 20 usec for direction setup and hold (20000)

Jitter will not cause loss of steps unless its so large that it causes steps in the
velocity during acceleration that the drive/motor cannot follow and result in a stall
(and this is not an issue with closed loop servos like the Clearpaths)

Any loss of steps in the step generator is detectable via the following error,
basically the way this works is that every servo period, LinuxCNC reads the step generators
step count (and fractional step count) and writes a new velocity to the step generator so that
any errors are corrected (via the PID component).

hmm interesting in pnconf I selected 'custom' drivers because Idk what to put, but I made sure to put 5k 5k 10k, but after you mention it the ini is clearly different.

just to confirm my settings changes in the ini should be:

# these are in nanoseconds
DIRSETUP = 20000
DIRHOLD = 20000
STEPLEN = 5000
STEPSPACE = 5000
STEP_SCALE = 4000
MIN_LIMIT = -91.0
MAX_LIMIT = 91.0
HOME_OFFSET = 0.0
# these servos are 800 steps/r and 5 Turns/1" ball screws step scale is really 4k

yes

thanks guys I will try this now.

vimeo.com/303568087

This worked!! I can now attempt a production run. I will try to pay it forward and make a tut of my endeavors problems and fixes. Thanks PCW and Todd for the fast help.