Re: [CAD_CAM_EDM_DRO] Re: backlash

At 23:47 04/01/2003 -0600, you wrote:

>Tony Jeffree wrote:
>
> >Tim -
> >
> >That is only because the crude approximations to backlash compensation
> >start with the assumptions that:
> >
> >(a) a given program will not do both climb milling and "normal" milling, and
> >
>No, the problem with climb milling is that the rotating cutting tool can
>pull the work into the
>cutter, causing a broken cutter, wrecked workpiece, damaged machine, etc.
>
> >(b) the backlash is constant over the travel of a given leadscrew, and
> >(c) the backlash value is accurately known.
> >
> >
>No, it still doesn't work. The problem is that when an axis reverses
>while the cutter is making a
>cut, the machine cannot get from one side of the backlash to the other
>side instantaneously, so it
>allows the table to be unconstrained by the motors for a moment. This
>is a situation whenever
>you are milling a circular hole or pocket, a wonderful thing to be able
>to do with a CNC machine.
>
> >It would be possible to provide sufficient information to the control
> >software to allow it to do a significantly better job of backlash
> >compensation than most controls manage currently; however, the real
> >question is whether it is a worthwhile thing to do, given the alternative
> >of building a better machine to start with.
> >
> >
>The problem stated above just can't be fixed with software, no matter
>how hard you try.

It could be made to work as long as the control software has a complete
picture of the backlash at all positions on each axis (not just the average
or max value), and as long as it is programmed to interrupt a cut (in order
to change the direction of pre-load) before the direction of cut changes.


Regards,
Tony