Dennis,
Some thoughts:
- backlash through the encoder – so when you go from forward to reverse there are a number of missed pulses.
- stepper missing steps – maybe at the point of reverse direction, maybe during the screwcutting run or when returning
- noise into the encoder path
- since you keep the halfnuts engaged, I don't think the issue can be the config settings since errors here should only result in incorrect pitches/TPIs and not a loss in sync
#1 would appear to be most consistent with the reported symptoms – missing pulses to lose the sync between the leadscrew and the mandrel.
#2 again, I can only see this related to the change of direction at each end of the run. Any inconsistency during the cutting run should be visible as a lack of parallelism between the first and second cuts.
#3 same as above…
I can run my ELS by turning the mandrel by hand so in your case, I would:
- put a long throw dti on the bed/saddle junction to see how much the saddle moved when I moved the mandrel by a full turn
- observe the effects of reversing direction
Both observations may shed more light on what is going on and turning the mandrel by hand will also eliminate your motor as a noise source.
Also, check that the various 0V connections are not linked – in particular, the Launchpad 0V and those for the stepper driver – including screen path connections.
Hope this helps, I have had my ELS running for a couple of weeks and it is really good once it is running.
Simon