EMC handling backlash with linear encoders and lookup tables

[Doug Fortune <pentam-at-home.com>]

Jon Elson wrote:

> Doug Fortune wrote:
>
> > > Why this happens is that when the table position needs to
> > > be moved just a hair the other way, the motor picks up
> > > speed crossing the backlash, while the encoder doesn't
> > > sense any movement.  Then, the motor takes up all the
> > > backlash, and the Acme screw suddenly moves the table
> > > too far before it can stop.  Then the process repeats in
> > > the other direction.
> >
> > Shouldn't this be built into EMC?  EMC should know
> > it takes X steps to go D distance, or in particular S steps
> > to go B (backlash distance).   It should also know
> > when it is reversing directions for that axis.
>
> Well, if the system uses open loop steppers, or encoders on the
> leadscrew, then EMC does this, but if the backlash is substantial,
> like a worn Acme screw, then the results are poor.  You are demanding
> the motor and leadscrew to move a finite distance in an infinitesimal
> amount of time, and that is not possible.  If the backlash is on the
> order of .001", like a slightly worn ballscrew, then it doesn't do too
> badly with it.  But, a worn Acme screw can have .015-.050" of
> backlash.  Even after adjusting the nuts, you may be stuck with .007"
> or so, due to greater wear in the center of the leadscrew itself.
> That's what I got on my old Bridgeport after fixing the nuts.

> > Yes, its a bit complicated, but isn't that supposed to
> > be the purpose of EMC knowing the backlash in each
> > of the axes?
>
> When the leadscrews are worn, the amount of backlash
> VARIES over the length of the screw!  It varied from less than
> .001" at the ends, to .007" or more at the middle, on mine.

Maybe once we find a reasonably priced linear encoder (which
of course directly senses table movement) then it will be
feasible to have EMC directly characterize the backlash
along the length of each axis (say every 1/10th, 1/100th or 1/1000")
and rather than a simple fixed value, the backlash is looked
up near that location on every reverse of an axis.

In fact, a slight extension of this characterization (say every
1/1000")  would even compensate automatically for the
non-linearity of the acme or ball screw.

After saving the lookup table, you  could even unmount the
linear encoder (and take it back to the rental store! :)    ).

Doug Fortune