Re: emc backlash compensation

[Fred Proctor <proctor-at-cme.nist.gov>]

Jon,

Backlash compensation worked for servo systems for a while, then I broke
it, then I fixed it in the March 15 release. I now have a Bridgeport
servo machine with about 0.003" of backlash on the Y axis due to a loose
belt, and I'll try backlash comp and see what happens.

Note that I changed the backlash comp algorithm that originally ran in
emc/src/emcmot/pid.c, so that it's done in emc/src/emcmot/emcmot.c. The
change is that the backlash is metered out upon a reversal according to
an acc-dec profile, rather than dumped all at once. This fixed a problem
with resonance on stepper systems.

I'll test on our Bridgeport and see what I come up with. If it works as
expected, then you'll want to get the latest release, although I know
you're loathe to jump off that December release.

--Fred